File openssl__chacha20_poly1305_draft_and_rfc_ossl102g.patch of Package nginx

File openssl__chacha20_poly1305_draft_and_rfc_ossl102g.patch of Package nginx (Revision dc32076a267a76349b0f645869df36af)

Currently displaying revision dc32076a267a76349b0f645869df36af, show latest

From d21c75c622eb13a80080ed15fa30474f806a0a0f Mon Sep 17 00:00:00 2001
From: Vlad Krasnov <vlad@cloudflare.com>
Date: Fri, 12 Feb 2016 18:25:11 -0800
Subject: [PATCH] Implementation of draft and RFC versions of CHACHA20-POLY1305
 ciphers

---
 Configure                                    |  48 +-
 Makefile.org                                 |   4 +-
 apps/speed.c                                 |  30 +-
 crypto/chacha20poly1305/Makefile             |  97 +++
 crypto/chacha20poly1305/asm/chacha20_avx.pl  | 408 +++++++++++
 crypto/chacha20poly1305/asm/chacha20_avx2.pl | 443 ++++++++++++
 crypto/chacha20poly1305/asm/poly1305_avx.pl  | 732 ++++++++++++++++++++
 crypto/chacha20poly1305/asm/poly1305_avx2.pl | 984 +++++++++++++++++++++++++++
 crypto/chacha20poly1305/asm/poly1305_x64.pl  | 281 ++++++++
 crypto/chacha20poly1305/chacha20.c           | 162 +++++
 crypto/chacha20poly1305/chacha20poly1305.h   |  79 +++
 crypto/chacha20poly1305/chapolytest.c        | 470 +++++++++++++
 crypto/chacha20poly1305/poly1305.c           | 287 ++++++++
 crypto/cryptlib.c                            |  10 -
 crypto/evp/Makefile                          |   7 +-
 crypto/evp/e_chacha20poly1305.c              | 435 ++++++++++++
 crypto/evp/evp.h                             |   4 +
 ssl/s3_lib.c                                 | 119 ++++
 ssl/ssl.h                                    |   2 +
 ssl/ssl_ciph.c                               |  60 +-
 ssl/ssl_locl.h                               |   2 +
 ssl/tls1.h                                   |  28 +
 test/Makefile                                |  17 +-
 23 files changed, 4655 insertions(+), 54 deletions(-)
 create mode 100644 crypto/chacha20poly1305/Makefile
 create mode 100644 crypto/chacha20poly1305/asm/chacha20_avx.pl
 create mode 100644 crypto/chacha20poly1305/asm/chacha20_avx2.pl
 create mode 100644 crypto/chacha20poly1305/asm/poly1305_avx.pl
 create mode 100644 crypto/chacha20poly1305/asm/poly1305_avx2.pl
 create mode 100644 crypto/chacha20poly1305/asm/poly1305_x64.pl
 create mode 100644 crypto/chacha20poly1305/chacha20.c
 create mode 100644 crypto/chacha20poly1305/chacha20poly1305.h
 create mode 100644 crypto/chacha20poly1305/chapolytest.c
 create mode 100644 crypto/chacha20poly1305/poly1305.c
 create mode 100644 crypto/evp/e_chacha20poly1305.c

diff --git a/Configure b/Configure
index 4a715dc..f3ab6cd 100755
--- a/Configure
+++ b/Configure
@@ -146,25 +146,25 @@ my $tlib="-lnsl -lsocket";
 my $bits1="THIRTY_TWO_BIT ";
 my $bits2="SIXTY_FOUR_BIT ";
 
-my $x86_asm="x86cpuid.o:bn-586.o co-586.o x86-mont.o x86-gf2m.o::des-586.o crypt586.o:aes-586.o vpaes-x86.o aesni-x86.o:bf-586.o:md5-586.o:sha1-586.o sha256-586.o sha512-586.o:cast-586.o:rc4-586.o:rmd-586.o:rc5-586.o:wp_block.o wp-mmx.o:cmll-x86.o:ghash-x86.o:";
+my $x86_asm="x86cpuid.o:bn-586.o co-586.o x86-mont.o x86-gf2m.o::des-586.o crypt586.o:aes-586.o vpaes-x86.o aesni-x86.o:bf-586.o:md5-586.o:sha1-586.o sha256-586.o sha512-586.o:cast-586.o:rc4-586.o:rmd-586.o:rc5-586.o:wp_block.o wp-mmx.o:cmll-x86.o:ghash-x86.o::";
 
 my $x86_elf_asm="$x86_asm:elf";
 
-my $x86_64_asm="x86_64cpuid.o:x86_64-gcc.o x86_64-mont.o x86_64-mont5.o x86_64-gf2m.o rsaz_exp.o rsaz-x86_64.o rsaz-avx2.o:ecp_nistz256.o ecp_nistz256-x86_64.o::aes-x86_64.o vpaes-x86_64.o bsaes-x86_64.o aesni-x86_64.o aesni-sha1-x86_64.o aesni-sha256-x86_64.o aesni-mb-x86_64.o::md5-x86_64.o:sha1-x86_64.o sha256-x86_64.o sha512-x86_64.o sha1-mb-x86_64.o sha256-mb-x86_64.o::rc4-x86_64.o rc4-md5-x86_64.o:::wp-x86_64.o:cmll-x86_64.o cmll_misc.o:ghash-x86_64.o aesni-gcm-x86_64.o:";
-my $ia64_asm="ia64cpuid.o:bn-ia64.o ia64-mont.o:::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o::rc4-ia64.o rc4_skey.o:::::ghash-ia64.o::void";
-my $sparcv9_asm="sparcv9cap.o sparccpuid.o:bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o sparct4-mont.o sparcv9-gf2m.o::des_enc-sparc.o fcrypt_b.o dest4-sparcv9.o:aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o::md5-sparcv9.o:sha1-sparcv9.o sha256-sparcv9.o sha512-sparcv9.o::::::camellia.o cmll_misc.o cmll_cbc.o cmllt4-sparcv9.o:ghash-sparcv9.o::void";
-my $sparcv8_asm=":sparcv8.o::des_enc-sparc.o fcrypt_b.o:::::::::::::void";
-my $alpha_asm="alphacpuid.o:bn_asm.o alpha-mont.o::::::sha1-alpha.o:::::::ghash-alpha.o::void";
-my $mips64_asm=":bn-mips.o mips-mont.o:::aes_cbc.o aes-mips.o:::sha1-mips.o sha256-mips.o sha512-mips.o::::::::";
+my $x86_64_asm="x86_64cpuid.o:x86_64-gcc.o x86_64-mont.o x86_64-mont5.o x86_64-gf2m.o rsaz_exp.o rsaz-x86_64.o rsaz-avx2.o:ecp_nistz256.o ecp_nistz256-x86_64.o::aes-x86_64.o vpaes-x86_64.o bsaes-x86_64.o aesni-x86_64.o aesni-sha1-x86_64.o aesni-sha256-x86_64.o aesni-mb-x86_64.o::md5-x86_64.o:sha1-x86_64.o sha256-x86_64.o sha512-x86_64.o sha1-mb-x86_64.o sha256-mb-x86_64.o::rc4-x86_64.o rc4-md5-x86_64.o:::wp-x86_64.o:cmll-x86_64.o cmll_misc.o:ghash-x86_64.o aesni-gcm-x86_64.o::chacha20_avx.o chacha20_avx2.o poly1305_x64.o poly1305_avx2.o";
+my $ia64_asm="ia64cpuid.o:bn-ia64.o ia64-mont.o:::aes_core.o aes_cbc.o aes-ia64.o::md5-ia64.o:sha1-ia64.o sha256-ia64.o sha512-ia64.o::rc4-ia64.o rc4_skey.o:::::ghash-ia64.o:::void";
+my $sparcv9_asm="sparcv9cap.o sparccpuid.o:bn-sparcv9.o sparcv9-mont.o sparcv9a-mont.o vis3-mont.o sparct4-mont.o sparcv9-gf2m.o::des_enc-sparc.o fcrypt_b.o dest4-sparcv9.o:aes_core.o aes_cbc.o aes-sparcv9.o aest4-sparcv9.o::md5-sparcv9.o:sha1-sparcv9.o sha256-sparcv9.o sha512-sparcv9.o::::::camellia.o cmll_misc.o cmll_cbc.o cmllt4-sparcv9.o:ghash-sparcv9.o:::void";
+my $sparcv8_asm=":sparcv8.o::des_enc-sparc.o fcrypt_b.o::::::::::::::void";
+my $alpha_asm="alphacpuid.o:bn_asm.o alpha-mont.o::::::sha1-alpha.o:::::::ghash-alpha.o:::void";
+my $mips64_asm=":bn-mips.o mips-mont.o:::aes_cbc.o aes-mips.o:::sha1-mips.o sha256-mips.o sha512-mips.o:::::::::";
 my $mips32_asm=$mips64_asm; $mips32_asm =~ s/\s*sha512\-mips\.o//;
-my $s390x_asm="s390xcap.o s390xcpuid.o:bn-s390x.o s390x-mont.o s390x-gf2m.o:::aes-s390x.o aes-ctr.o aes-xts.o:::sha1-s390x.o sha256-s390x.o sha512-s390x.o::rc4-s390x.o:::::ghash-s390x.o:";
-my $armv4_asm="armcap.o armv4cpuid.o:bn_asm.o armv4-mont.o armv4-gf2m.o:::aes_cbc.o aes-armv4.o bsaes-armv7.o aesv8-armx.o:::sha1-armv4-large.o sha256-armv4.o sha512-armv4.o:::::::ghash-armv4.o ghashv8-armx.o::void";
-my $aarch64_asm="armcap.o arm64cpuid.o mem_clr.o::::aes_core.o aes_cbc.o aesv8-armx.o:::sha1-armv8.o sha256-armv8.o sha512-armv8.o:::::::ghashv8-armx.o:";
-my $parisc11_asm="pariscid.o:bn_asm.o parisc-mont.o:::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o::32";
-my $parisc20_asm="pariscid.o:pa-risc2W.o parisc-mont.o:::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o::64";
-my $ppc64_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o:::aes_core.o aes_cbc.o aes-ppc.o vpaes-ppc.o aesp8-ppc.o:::sha1-ppc.o sha256-ppc.o sha512-ppc.o sha256p8-ppc.o sha512p8-ppc.o:::::::ghashp8-ppc.o:";
+my $s390x_asm="s390xcap.o s390xcpuid.o:bn-s390x.o s390x-mont.o s390x-gf2m.o:::aes-s390x.o aes-ctr.o aes-xts.o:::sha1-s390x.o sha256-s390x.o sha512-s390x.o::rc4-s390x.o:::::ghash-s390x.o::";
+my $armv4_asm="armcap.o armv4cpuid.o:bn_asm.o armv4-mont.o armv4-gf2m.o:::aes_cbc.o aes-armv4.o bsaes-armv7.o aesv8-armx.o:::sha1-armv4-large.o sha256-armv4.o sha512-armv4.o:::::::ghash-armv4.o ghashv8-armx.o:::void";
+my $aarch64_asm="armcap.o arm64cpuid.o mem_clr.o::::aes_core.o aes_cbc.o aesv8-armx.o:::sha1-armv8.o sha256-armv8.o sha512-armv8.o:::::::ghashv8-armx.o::";
+my $parisc11_asm="pariscid.o:bn_asm.o parisc-mont.o:::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o:::32";
+my $parisc20_asm="pariscid.o:pa-risc2W.o parisc-mont.o:::aes_core.o aes_cbc.o aes-parisc.o:::sha1-parisc.o sha256-parisc.o sha512-parisc.o::rc4-parisc.o:::::ghash-parisc.o:::64";
+my $ppc64_asm="ppccpuid.o ppccap.o:bn-ppc.o ppc-mont.o ppc64-mont.o:::aes_core.o aes_cbc.o aes-ppc.o vpaes-ppc.o aesp8-ppc.o:::sha1-ppc.o sha256-ppc.o sha512-ppc.o sha256p8-ppc.o sha512p8-ppc.o:::::::ghashp8-ppc.o::";
 my $ppc32_asm=$ppc64_asm;
-my $no_asm="::::::::::::::::void";
+my $no_asm=":::::::::::::::::void";
 
 # As for $BSDthreads. Idea is to maintain "collective" set of flags,
 # which would cover all BSD flavors. -pthread applies to them all, 
@@ -710,6 +710,7 @@ my $idx_wp_obj = $idx++;
 my $idx_cmll_obj = $idx++;
 my $idx_modes_obj = $idx++;
 my $idx_engines_obj = $idx++;
+my $idx_chapoly_obj = $idx++;
 my $idx_perlasm_scheme = $idx++;
 my $idx_dso_scheme = $idx++;
 my $idx_shared_target = $idx++;
@@ -752,6 +753,7 @@ my $bf	="crypto/bf/bf_locl.h";
 my $bn_asm	="bn_asm.o";
 my $des_enc="des_enc.o fcrypt_b.o";
 my $aes_enc="aes_core.o aes_cbc.o";
+my $chapoly_enc="";
 my $bf_enc	="bf_enc.o";
 my $cast_enc="c_enc.o";
 my $rc4_enc="rc4_enc.o rc4_skey.o";
@@ -1210,7 +1212,7 @@ $openssldir=$prefix . "/" . $openssldir if $openssldir !~ /(^\/|^[a-zA-Z]:[\\\/]
 
 print "IsMK1MF=$IsMK1MF\n";
 
-my @fields = split(/\s*:\s*/,$table{$target} . ":" x 30 , -1);
+my @fields = split(/\s*:\s*/,$table{$target} . ":" x 31 , -1);
 my $cc = $fields[$idx_cc];
 # Allow environment CC to override compiler...
 if($ENV{CC}) {
@@ -1239,6 +1241,7 @@ my $wp_obj = $fields[$idx_wp_obj];
 my $cmll_obj = $fields[$idx_cmll_obj];
 my $modes_obj = $fields[$idx_modes_obj];
 my $engines_obj = $fields[$idx_engines_obj];
+my $chapoly_obj = $fields[$idx_chapoly_obj];
 my $perlasm_scheme = $fields[$idx_perlasm_scheme];
 my $dso_scheme = $fields[$idx_dso_scheme];
 my $shared_target = $fields[$idx_shared_target];
@@ -1405,7 +1408,7 @@ if ($no_asm)
 	{
 	$cpuid_obj=$bn_obj=$ec_obj=
 	$des_obj=$aes_obj=$bf_obj=$cast_obj=$rc4_obj=$rc5_obj=$cmll_obj=
-	$modes_obj=$sha1_obj=$md5_obj=$rmd160_obj=$wp_obj=$engines_obj="";
+	$modes_obj=$sha1_obj=$md5_obj=$rmd160_obj=$wp_obj=$engines_obj=$chapoly_obj="";
 	}
 
 if (!$no_shared)
@@ -1558,6 +1561,14 @@ $bf_obj=$bf_enc		unless ($bf_obj =~ /\.o$/);
 $cast_obj=$cast_enc	unless ($cast_obj =~ /\.o$/);
 $rc4_obj=$rc4_enc	unless ($rc4_obj =~ /\.o$/);
 $rc5_obj=$rc5_enc	unless ($rc5_obj =~ /\.o$/);
+if ($chapoly_obj =~ /\.o$/)
+	{
+	$cflags.=" -DCHAPOLY_x86_64_ASM";
+	}
+else
+	{
+	$chapoly_obj=$chapoly_enc;
+	}
 if ($sha1_obj =~ /\.o$/)
 	{
 #	$sha1_obj=$sha1_enc;
@@ -1740,6 +1751,7 @@ while (<IN>)
 	s/^WP_ASM_OBJ=.*$/WP_ASM_OBJ= $wp_obj/;
 	s/^CMLL_ENC=.*$/CMLL_ENC= $cmll_obj/;
 	s/^MODES_ASM_OBJ.=*$/MODES_ASM_OBJ= $modes_obj/;
+	s/^CHAPOLY_ENC=.*$/CHAPOLY_ENC= $chapoly_obj/;
 	s/^ENGINES_ASM_OBJ.=*$/ENGINES_ASM_OBJ= $engines_obj/;
 	s/^PERLASM_SCHEME=.*$/PERLASM_SCHEME= $perlasm_scheme/;
 	s/^PROCESSOR=.*/PROCESSOR= $processor/;
@@ -1802,6 +1814,7 @@ print "RMD160_OBJ_ASM=$rmd160_obj\n";
 print "CMLL_ENC      =$cmll_obj\n";
 print "MODES_OBJ     =$modes_obj\n";
 print "ENGINES_OBJ   =$engines_obj\n";
+print "CHAPOLY_ENC   =$chapoly_obj\n";
 print "PROCESSOR     =$processor\n";
 print "RANLIB        =$ranlib\n";
 print "ARFLAGS       =$arflags\n";
@@ -2200,7 +2213,7 @@ sub print_table_entry
 	my ($cc, $cflags, $unistd, $thread_cflag, $sys_id, $lflags,
 	    $bn_ops, $cpuid_obj, $bn_obj, $ec_obj, $des_obj, $aes_obj, $bf_obj,
 	    $md5_obj, $sha1_obj, $cast_obj, $rc4_obj, $rmd160_obj,
-	    $rc5_obj, $wp_obj, $cmll_obj, $modes_obj, $engines_obj,
+	    $rc5_obj, $wp_obj, $cmll_obj, $modes_obj, $engines_obj, $chapoly_obj,
 	    $perlasm_scheme, $dso_scheme, $shared_target, $shared_cflag,
 	    $shared_ldflag, $shared_extension, $ranlib, $arflags, $multilib)=
 	split(/\s*:\s*/,$table{$target} . ":" x 30 , -1);
@@ -2231,6 +2244,7 @@ sub print_table_entry
 \$cmll_obj     = $cmll_obj
 \$modes_obj    = $modes_obj
 \$engines_obj  = $engines_obj
+\$chapoly_obj  = $chapoly_obj
 \$perlasm_scheme = $perlasm_scheme
 \$dso_scheme   = $dso_scheme
 \$shared_target= $shared_target
diff --git a/Makefile.org b/Makefile.org
index 76fdbdf..6556ef6 100644
--- a/Makefile.org
+++ b/Makefile.org
@@ -91,6 +91,7 @@ BN_ASM= bn_asm.o
 EC_ASM=
 DES_ENC= des_enc.o fcrypt_b.o
 AES_ENC= aes_core.o aes_cbc.o
+CHAPOLY_ENC=
 BF_ENC= bf_enc.o
 CAST_ENC= c_enc.o
 RC4_ENC= rc4_enc.o
@@ -148,7 +149,7 @@ SDIRS=  \
 	bn ec rsa dsa ecdsa dh ecdh dso engine \
 	buffer bio stack lhash rand err \
 	evp asn1 pem x509 x509v3 conf txt_db pkcs7 pkcs12 comp ocsp ui krb5 \
-	cms pqueue ts jpake srp store cmac
+	cms pqueue ts jpake srp store cmac chacha20poly1305
 # keep in mind that the above list is adjusted by ./Configure
 # according to no-xxx arguments...
 
@@ -234,6 +235,7 @@ BUILDENV=	LC_ALL=C PLATFORM='$(PLATFORM)' PROCESSOR='$(PROCESSOR)'\
 		WP_ASM_OBJ='$(WP_ASM_OBJ)'			\
 		MODES_ASM_OBJ='$(MODES_ASM_OBJ)'		\
 		ENGINES_ASM_OBJ='$(ENGINES_ASM_OBJ)'		\
+		CHAPOLY_ENC='$(CHAPOLY_ENC)'			\
 		PERLASM_SCHEME='$(PERLASM_SCHEME)'		\
 		FIPSLIBDIR='${FIPSLIBDIR}'			\
 		FIPSDIR='${FIPSDIR}'				\
diff --git a/apps/speed.c b/apps/speed.c
index 95adcc1..94b80a1 100644
--- a/apps/speed.c
+++ b/apps/speed.c
@@ -1,4 +1,4 @@
-/* apps/speed.c */
+/* apps/speed.c -*- mode:C; c-file-style: "eay" -*- */
 /* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
  * All rights reserved.
  *
@@ -226,7 +226,7 @@
 # endif
 
 # undef BUFSIZE
-# define BUFSIZE ((long)1024*8+1)
+# define BUFSIZE ((long)1024*8+16)
 static volatile int run = 0;
 
 static int mr = 0;
@@ -241,7 +241,7 @@ static void print_result(int alg, int run_no, int count, double time_used);
 static int do_multi(int multi);
 # endif
 
-# define ALGOR_NUM       30
+# define ALGOR_NUM       31
 # define SIZE_NUM        5
 # define RSA_NUM         4
 # define DSA_NUM         3
@@ -256,7 +256,7 @@ static const char *names[ALGOR_NUM] = {
     "aes-128 cbc", "aes-192 cbc", "aes-256 cbc",
     "camellia-128 cbc", "camellia-192 cbc", "camellia-256 cbc",
     "evp", "sha256", "sha512", "whirlpool",
-    "aes-128 ige", "aes-192 ige", "aes-256 ige", "ghash"
+    "aes-128 ige", "aes-192 ige", "aes-256 ige", "ghash", "chacha20-poly1305"
 };
 
 static double results[ALGOR_NUM][SIZE_NUM];
@@ -516,6 +516,7 @@ int MAIN(int argc, char **argv)
 # define D_IGE_192_AES   27
 # define D_IGE_256_AES   28
 # define D_GHASH         29
+# define D_CHAPOLY       30
     double d = 0.0;
     long c[ALGOR_NUM][SIZE_NUM];
 # define R_DSA_512       0
@@ -972,6 +973,9 @@ int MAIN(int argc, char **argv)
             doit[D_CBC_256_CML] = 1;
         } else
 # endif
+        if (strcmp(*argv, "chacha20-poly1305") == 0) {
+            doit[D_CHAPOLY] = 1;
+        } else
 # ifndef OPENSSL_NO_RSA
         if (strcmp(*argv, "rsa") == 0) {
             rsa_doit[R_RSA_512] = 1;
@@ -1139,6 +1143,7 @@ int MAIN(int argc, char **argv)
             BIO_printf(bio_err, "rc4");
 # endif
             BIO_printf(bio_err, "\n");
+            BIO_printf(bio_err, "chacha20-poly1305\n");
 
 # ifndef OPENSSL_NO_RSA
             BIO_printf(bio_err, "rsa512   rsa1024  rsa2048  rsa4096\n");
@@ -1370,6 +1375,7 @@ int MAIN(int argc, char **argv)
     c[D_IGE_192_AES][0] = count;
     c[D_IGE_256_AES][0] = count;
     c[D_GHASH][0] = count;
+    c[D_CHAPOLY][0] = count;
 
     for (i = 1; i < SIZE_NUM; i++) {
         c[D_MD2][i] = c[D_MD2][0] * 4 * lengths[0] / lengths[i];
@@ -1821,6 +1827,22 @@ int MAIN(int argc, char **argv)
         CRYPTO_gcm128_release(ctx);
     }
 # endif
+    if (doit[D_CHAPOLY]) {
+        EVP_CIPHER_CTX ctx;
+        EVP_CIPHER_CTX_init(&ctx);
+        EVP_CipherInit_ex(&ctx, EVP_chacha20_poly1305(), NULL, key32, iv, 1);
+
+        for (j = 0; j < SIZE_NUM; j++) {
+             print_message(names[D_CHAPOLY], c[D_CHAPOLY][j], lengths[j]);
+             Time_F(START);
+             for (count = 0, run = 1; COND(c[D_CHAPOLY][j]); count++) {
+                  EVP_CIPHER_CTX_ctrl(&ctx, EVP_CTRL_AEAD_TLS1_AAD, 13, buf);
+                  EVP_Cipher(&ctx, buf, buf, (unsigned long)lengths[j] + 16);
+             }
+             d = Time_F(STOP);
+             print_result(D_CHAPOLY, j, count, d);
+        }
+    }
 # ifndef OPENSSL_NO_CAMELLIA
     if (doit[D_CBC_128_CML]) {
         for (j = 0; j < SIZE_NUM; j++) {
diff --git a/crypto/chacha20poly1305/Makefile b/crypto/chacha20poly1305/Makefile
new file mode 100644
index 0000000..446eb27
--- /dev/null
+++ b/crypto/chacha20poly1305/Makefile
@@ -0,0 +1,97 @@
+#
+# crypto/chacha20poly1305/Makefile
+#
+
+DIR=	chacha20poly1305
+TOP=	../..
+CC=	cc
+CPP=	$(CC) -E
+INCLUDES=
+CFLAG=-g
+MAKEFILE=	Makefile
+AR=		ar r
+
+CHAPOLY_ENC=
+
+CFLAGS= $(INCLUDES) $(CFLAG)
+ASFLAGS= $(INCLUDES) $(ASFLAG)
+AFLAGS= $(ASFLAGS)
+
+GENERAL=Makefile
+TEST=chapolytest.c
+APPS=
+
+LIB=$(TOP)/libcrypto.a
+LIBSRC=chacha20.c poly1305.c
+LIBOBJ=chacha20.o poly1305.o $(CHAPOLY_ENC)
+
+SRC= $(LIBSRC)
+
+EXHEADER=chacha20poly1305.h
+HEADER= $(EXHEADER)
+
+ALL=    $(GENERAL) $(SRC) $(HEADER)
+
+top:
+	(cd ../..; $(MAKE) DIRS=crypto SDIRS=$(DIR) sub_all)
+
+all:	lib
+
+lib:	$(LIBOBJ)
+	$(AR) $(LIB) $(LIBOBJ)
+	$(RANLIB) $(LIB) || echo Never mind.
+	@touch lib
+
+poly1305_x64.s:asm/poly1305_x64.pl
+	$(PERL) asm/poly1305_x64.pl $(PERLASM_SCHEME) > $@
+chacha20_avx.s:asm/chacha20_avx.pl
+	$(PERL) asm/chacha20_avx.pl $(PERLASM_SCHEME) > $@
+poly1305_avx.s:asm/poly1305_avx.pl
+	$(PERL) asm/poly1305_avx.pl $(PERLASM_SCHEME) > $@
+chacha20_avx2.s:asm/chacha20_avx2.pl
+	$(PERL) asm/chacha20_avx2.pl $(PERLASM_SCHEME) > $@
+poly1305_avx2.s:asm/poly1305_avx2.pl
+	$(PERL) asm/poly1305_avx2.pl $(PERLASM_SCHEME) > $@
+
+files:
+	$(PERL) $(TOP)/util/files.pl Makefile >> $(TOP)/MINFO
+
+links:
+	@$(PERL) $(TOP)/util/mklink.pl ../../include/openssl $(EXHEADER)
+	@$(PERL) $(TOP)/util/mklink.pl ../../test $(TEST)
+	@$(PERL) $(TOP)/util/mklink.pl ../../apps $(APPS)
+
+install:
+	@[ -n "$(INSTALLTOP)" ] # should be set by top Makefile...
+	@headerlist="$(EXHEADER)"; for i in $$headerlist ; \
+	do  \
+	(cp $$i $(INSTALL_PREFIX)$(INSTALLTOP)/include/openssl/$$i; \
+	chmod 644 $(INSTALL_PREFIX)$(INSTALLTOP)/include/openssl/$$i ); \
+	done;
+
+tags:
+	ctags $(SRC)
+
+tests:
+
+chapolytest: top chapolytest.c $(LIB)
+	$(CC) $(CFLAGS) -Wall -Werror -g -o chapolytest cahpolytest.c $(LIB)
+
+lint:
+	lint -DLINT $(INCLUDES) $(SRC)>fluff
+
+depend:
+	@[ -n "$(MAKEDEPEND)" ] # should be set by upper Makefile...
+	$(MAKEDEPEND) -- $(CFLAG) $(INCLUDES) $(DEPFLAG) -- $(PROGS) $(LIBSRC)
+
+dclean:
+	$(PERL) -pe 'if (/^# DO NOT DELETE THIS LINE/) {print; exit(0);}' $(MAKEFILE) >Makefile.new
+	mv -f Makefile.new $(MAKEFILE)
+
+clean:
+	rm -f *.s *.o *.obj lib tags core .pure .nfs* *.old *.bak fluff
+
+# DO NOT DELETE THIS LINE -- make depend depends on it.
+
+chacha20.o: ../../include/openssl/chacha20poly1305.h chacha20.c
+poly1305.o: ../../include/openssl/chacha20poly1305.h poly1305.c
diff --git a/crypto/chacha20poly1305/asm/chacha20_avx.pl b/crypto/chacha20poly1305/asm/chacha20_avx.pl
new file mode 100644
index 0000000..bf3e3f0
--- /dev/null
+++ b/crypto/chacha20poly1305/asm/chacha20_avx.pl
@@ -0,0 +1,408 @@
+#!/usr/bin/env perl
+
+##############################################################################
+#                                                                            #
+# Copyright 2014 Intel Corporation                                           #
+#                                                                            #
+# Licensed under the Apache License, Version 2.0 (the "License");            #
+# you may not use this file except in compliance with the License.           #
+# You may obtain a copy of the License at                                    #
+#                                                                            #
+#    http://www.apache.org/licenses/LICENSE-2.0                              #
+#                                                                            #
+# Unless required by applicable law or agreed to in writing, software        #
+# distributed under the License is distributed on an "AS IS" BASIS,          #
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.   #
+# See the License for the specific language governing permissions and        #
+# limitations under the License.                                             #
+#                                                                            #
+##############################################################################
+#                                                                            #
+#  Developers and authors:                                                   #
+#  Shay Gueron (1, 2), and Vlad Krasnov (1)                                  #
+#  (1) Intel Corporation, Israel Development Center                          #
+#  (2) University of Haifa                                                   #
+#                                                                            #
+# Related work:                                                              #
+# M. Goll, S. Gueron, "Vectorization on ChaCha Stream Cipher", IEEE          #
+#          Proceedings of 11th International Conference on Information       #
+#          Technology: New Generations (ITNG 2014), 612-615 (2014).          #
+# M. Goll, S. Gueron, "Vectorization on Poly1305 Message Authentication Code"#
+#           to be published.                                                 #
+# A. Langley, chacha20poly1305 for the AEAD head                             #
+# https://git.openssl.org/gitweb/?p=openssl.git;a=commit;h=9a8646510b3d0a48e950748f7a2aaa12ed40d5e0  #
+##############################################################################
+
+
+$flavour = shift;
+$output  = shift;
+if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+
+$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
+( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
+die "can't locate x86_64-xlate.pl";
+
+open OUT,"| \"$^X\" $xlate $flavour $output";
+*STDOUT=*OUT;
+
+if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
+    =~ /GNU assembler version ([2-9]\.[0-9]+)/) {
+  $avx = ($1>=2.19) + ($1>=2.22);
+}
+
+if ($win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
+      `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+  $avx = ($1>=2.09) + ($1>=2.10);
+}
+
+if ($win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
+      `ml64 2>&1` =~ /Version ([0-9]+)\./) {
+  $avx = ($1>=10) + ($1>=11);
+}
+
+if (`$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9])\.([0-9]+)/) {
+  my $ver = $2 + $3/100.0;  # 3.1->3.01, 3.10->3.10
+  $avx = ($ver>=3.0) + ($ver>=3.01);
+}
+
+if ($avx>=1) {{
+
+my ($rol8, $rol16, $state_cdef, $tmp,
+    $v0, $v1, $v2, $v3, $v4, $v5, $v6, $v7,
+    $v8, $v9, $v10, $v11)=map("%xmm$_",(0..15));
+
+sub chacha_qr {
+
+my ($a,$b,$c,$d)=@_;
+$code.=<<___;
+
+  vpaddd  $b, $a, $a            # a += b
+  vpxor   $a, $d, $d            # d ^= a
+  vpshufb $rol16, $d, $d        # d <<<= 16
+
+  vpaddd  $d, $c, $c            # c += d
+  vpxor   $c, $b, $b            # b ^= c
+  vpslld  \$12, $b, $tmp
+  vpsrld  \$20, $b, $b
+  vpxor   $tmp, $b, $b          # b <<<= 12
+
+  vpaddd  $b, $a, $a            # a += b
+  vpxor   $a, $d, $d            # d ^= a
+  vpshufb $rol8, $d, $d         # d <<<= 8
+
+  vpaddd  $d, $c, $c            # c += d
+  vpxor   $c, $b, $b            # b ^= c
+
+  vpslld  \$7, $b, $tmp
+  vpsrld  \$25, $b, $b
+  vpxor   $tmp, $b, $b          # b <<<= 7
+___
+
+}
+
+$code.=<<___;
+
+.text
+.align 16
+chacha20_consts:
+.byte 'e','x','p','a','n','d',' ','3','2','-','b','y','t','e',' ','k'
+.rol8:
+.byte 3,0,1,2, 7,4,5,6, 11,8,9,10, 15,12,13,14
+.rol16:
+.byte 2,3,0,1, 6,7,4,5, 10,11,8,9, 14,15,12,13
+.avxInc:
+.quad 1,0
+
+___
+
+{
+
+
+my ($out, $in, $in_len, $key_ptr, $nr)
+   =("%rdi", "%rsi", "%rdx", "%rcx", "%r8");
+
+$code.=<<___;
+.globl chacha_20_core_avx
+.type  chacha_20_core_avx ,\@function,2
+.align 64
+chacha_20_core_avx:
+  vzeroupper
+
+  # Init state
+  vmovdqa  .rol8(%rip), $rol8
+  vmovdqa  .rol16(%rip), $rol16
+  vmovdqu  16*2($key_ptr), $state_cdef
+
+2:
+  cmp  \$64*3, $in_len
+  jb   2f
+
+  vmovdqa  chacha20_consts(%rip), $v0
+  vmovdqu  16*0($key_ptr), $v1
+  vmovdqu  16*1($key_ptr), $v2
+  vmovdqa  $state_cdef, $v3
+
+  vmovdqa  $v0, $v4
+  vmovdqa  $v0, $v8
+
+  vmovdqa  $v1, $v5
+  vmovdqa  $v1, $v9
+
+  vmovdqa  $v2, $v6
+  vmovdqa  $v2, $v10
+
+  vpaddq   .avxInc(%rip), $v3, $v7
+  vpaddq   .avxInc(%rip), $v7, $v11
+
+  mov  \$10, $nr
+
+  1:
+___
+
+    &chacha_qr( $v0, $v1, $v2, $v3);
+    &chacha_qr( $v4, $v5, $v6, $v7);
+    &chacha_qr( $v8, $v9,$v10,$v11);
+
+$code.=<<___;
+    vpalignr  \$4,  $v1,  $v1,  $v1
+    vpalignr  \$8,  $v2,  $v2,  $v2
+    vpalignr \$12,  $v3,  $v3,  $v3
+    vpalignr  \$4,  $v5,  $v5,  $v5
+    vpalignr  \$8,  $v6,  $v6,  $v6
+    vpalignr \$12,  $v7,  $v7,  $v7
+    vpalignr  \$4,  $v9,  $v9,  $v9
+    vpalignr  \$8, $v10, $v10, $v10
+    vpalignr \$12, $v11, $v11, $v11
+___
+
+    &chacha_qr( $v0, $v1, $v2, $v3);
+    &chacha_qr( $v4, $v5, $v6, $v7);
+    &chacha_qr( $v8, $v9,$v10,$v11);
+
+$code.=<<___;
+    vpalignr \$12,  $v1,  $v1,  $v1
+    vpalignr  \$8,  $v2,  $v2,  $v2
+    vpalignr  \$4,  $v3,  $v3,  $v3
+    vpalignr \$12,  $v5,  $v5,  $v5
+    vpalignr  \$8,  $v6,  $v6,  $v6
+    vpalignr  \$4,  $v7,  $v7,  $v7
+    vpalignr \$12,  $v9,  $v9,  $v9
+    vpalignr  \$8, $v10, $v10, $v10
+    vpalignr  \$4, $v11, $v11, $v11
+
+    dec  $nr
+
+  jnz  1b
+
+  vpaddd  chacha20_consts(%rip), $v0, $v0
+  vpaddd  chacha20_consts(%rip), $v4, $v4
+  vpaddd  chacha20_consts(%rip), $v8, $v8
+
+  vpaddd  16*0($key_ptr), $v1, $v1
+  vpaddd  16*0($key_ptr), $v5, $v5
+  vpaddd  16*0($key_ptr), $v9, $v9
+
+  vpaddd  16*1($key_ptr), $v2, $v2
+  vpaddd  16*1($key_ptr), $v6, $v6
+  vpaddd  16*1($key_ptr), $v10, $v10
+
+  vpaddd  $state_cdef, $v3, $v3
+  vpaddq  .avxInc(%rip), $state_cdef, $state_cdef
+  vpaddd  $state_cdef, $v7, $v7
+  vpaddq  .avxInc(%rip), $state_cdef, $state_cdef
+  vpaddd  $state_cdef, $v11, $v11
+  vpaddq  .avxInc(%rip), $state_cdef, $state_cdef
+
+  vpxor  16*0($in), $v0, $v0
+  vpxor  16*1($in), $v1, $v1
+  vpxor  16*2($in), $v2, $v2
+  vpxor  16*3($in), $v3, $v3
+
+  vmovdqu  $v0, 16*0($out)
+  vmovdqu  $v1, 16*1($out)
+  vmovdqu  $v2, 16*2($out)
+  vmovdqu  $v3, 16*3($out)
+
+  vpxor  16*4($in), $v4, $v4
+  vpxor  16*5($in), $v5, $v5
+  vpxor  16*6($in), $v6, $v6
+  vpxor  16*7($in), $v7, $v7
+
+  vmovdqu  $v4, 16*4($out)
+  vmovdqu  $v5, 16*5($out)
+  vmovdqu  $v6, 16*6($out)
+  vmovdqu  $v7, 16*7($out)
+
+  vpxor  16*8($in), $v8, $v8
+  vpxor  16*9($in), $v9, $v9
+  vpxor  16*10($in), $v10, $v10
+  vpxor  16*11($in), $v11, $v11
+
+  vmovdqu  $v8, 16*8($out)
+  vmovdqu  $v9, 16*9($out)
+  vmovdqu  $v10, 16*10($out)
+  vmovdqu  $v11, 16*11($out)
+
+  lea  16*12($in), $in
+  lea  16*12($out), $out
+  sub  \$16*12, $in_len
+
+  jmp  2b
+
+2:
+  cmp  \$64*2, $in_len
+  jb   2f
+
+  vmovdqa  chacha20_consts(%rip), $v0
+  vmovdqa  chacha20_consts(%rip), $v4
+  vmovdqu  16*0($key_ptr), $v1
+  vmovdqu  16*0($key_ptr), $v5
+  vmovdqu  16*1($key_ptr), $v2
+  vmovdqu  16*1($key_ptr), $v6
+  vmovdqu  16*1($key_ptr), $v10
+  vmovdqa  $state_cdef, $v3
+  vpaddq   .avxInc(%rip), $v3, $v7
+
+  mov  \$10, $nr
+
+  1:
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+    &chacha_qr($v4,$v5,$v6,$v7);
+
+$code.=<<___;
+    vpalignr  \$4, $v1, $v1, $v1
+    vpalignr  \$8, $v2, $v2, $v2
+    vpalignr \$12, $v3, $v3, $v3
+    vpalignr  \$4, $v5, $v5, $v5
+    vpalignr  \$8, $v6, $v6, $v6
+    vpalignr \$12, $v7, $v7, $v7
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+    &chacha_qr($v4,$v5,$v6,$v7);
+
+$code.=<<___;
+    vpalignr \$12, $v1, $v1, $v1
+    vpalignr  \$8, $v2, $v2, $v2
+    vpalignr  \$4, $v3, $v3, $v3
+    vpalignr \$12, $v5, $v5, $v5
+    vpalignr  \$8, $v6, $v6, $v6
+    vpalignr  \$4, $v7, $v7, $v7
+
+    dec  $nr
+
+  jnz  1b
+
+  vpaddd  chacha20_consts(%rip), $v0, $v0
+  vpaddd  chacha20_consts(%rip), $v4, $v4
+
+  vpaddd  16*0($key_ptr), $v1, $v1
+  vpaddd  16*0($key_ptr), $v5, $v5
+
+  vpaddd  16*1($key_ptr), $v2, $v2
+  vpaddd  16*1($key_ptr), $v6, $v6
+
+  vpaddd  $state_cdef, $v3, $v3
+  vpaddq  .avxInc(%rip), $state_cdef, $state_cdef
+  vpaddd  $state_cdef, $v7, $v7
+  vpaddq  .avxInc(%rip), $state_cdef, $state_cdef
+
+  vpxor  16*0($in), $v0, $v0
+  vpxor  16*1($in), $v1, $v1
+  vpxor  16*2($in), $v2, $v2
+  vpxor  16*3($in), $v3, $v3
+
+  vmovdqu  $v0, 16*0($out)
+  vmovdqu  $v1, 16*1($out)
+  vmovdqu  $v2, 16*2($out)
+  vmovdqu  $v3, 16*3($out)
+
+  vpxor  16*4($in), $v4, $v4
+  vpxor  16*5($in), $v5, $v5
+  vpxor  16*6($in), $v6, $v6
+  vpxor  16*7($in), $v7, $v7
+
+  vmovdqu  $v4, 16*4($out)
+  vmovdqu  $v5, 16*5($out)
+  vmovdqu  $v6, 16*6($out)
+  vmovdqu  $v7, 16*7($out)
+
+  lea  16*8($in), $in
+  lea  16*8($out), $out
+  sub  \$16*8, $in_len
+
+  jmp  2b
+2:
+  cmp  \$64, $in_len
+  jb  2f
+
+  vmovdqa  chacha20_consts(%rip), $v0
+  vmovdqu  16*0($key_ptr), $v1
+  vmovdqu  16*1($key_ptr), $v2
+  vmovdqa  $state_cdef, $v3
+
+  mov  \$10, $nr
+
+  1:
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+
+$code.=<<___;
+    vpalignr   \$4, $v1, $v1, $v1
+    vpalignr   \$8, $v2, $v2, $v2
+    vpalignr  \$12, $v3, $v3, $v3
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+$code.=<<___;
+    vpalignr  \$12, $v1, $v1, $v1
+    vpalignr   \$8, $v2, $v2, $v2
+    vpalignr   \$4, $v3, $v3, $v3
+
+    dec  $nr
+  jnz  1b
+
+  vpaddd  chacha20_consts(%rip), $v0, $v0
+  vpaddd  16*0($key_ptr), $v1, $v1
+  vpaddd  16*1($key_ptr), $v2, $v2
+  vpaddd  $state_cdef, $v3, $v3
+  vpaddq  .avxInc(%rip), $state_cdef, $state_cdef
+
+  vpxor  16*0($in), $v0, $v0
+  vpxor  16*1($in), $v1, $v1
+  vpxor  16*2($in), $v2, $v2
+  vpxor  16*3($in), $v3, $v3
+
+  vmovdqu  $v0, 16*0($out)
+  vmovdqu  $v1, 16*1($out)
+  vmovdqu  $v2, 16*2($out)
+  vmovdqu  $v3, 16*3($out)
+
+  lea  16*4($in), $in
+  lea  16*4($out), $out
+  sub  \$16*4, $in_len
+  jmp  2b
+
+2:
+  vmovdqu  $state_cdef, 16*2($key_ptr)
+
+  vzeroupper
+  ret
+.size  chacha_20_core_avx,.-chacha_20_core_avx
+___
+}
+}}
+
+
+$code =~ s/\`([^\`]*)\`/eval($1)/gem;
+
+print $code;
+
+close STDOUT;
+
diff --git a/crypto/chacha20poly1305/asm/chacha20_avx2.pl b/crypto/chacha20poly1305/asm/chacha20_avx2.pl
new file mode 100644
index 0000000..9f31f86
--- /dev/null
+++ b/crypto/chacha20poly1305/asm/chacha20_avx2.pl
@@ -0,0 +1,443 @@
+#!/usr/bin/env perl
+
+##############################################################################
+#                                                                            #
+# Copyright 2014 Intel Corporation                                           #
+#                                                                            #
+# Licensed under the Apache License, Version 2.0 (the "License");            #
+# you may not use this file except in compliance with the License.           #
+# You may obtain a copy of the License at                                    #
+#                                                                            #
+#    http://www.apache.org/licenses/LICENSE-2.0                              #
+#                                                                            #
+# Unless required by applicable law or agreed to in writing, software        #
+# distributed under the License is distributed on an "AS IS" BASIS,          #
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.   #
+# See the License for the specific language governing permissions and        #
+# limitations under the License.                                             #
+#                                                                            #
+##############################################################################
+#                                                                            #
+#  Developers and authors:                                                   #
+#  Shay Gueron (1, 2), and Vlad Krasnov (1)                                  #
+#  (1) Intel Corporation, Israel Development Center                          #
+#  (2) University of Haifa                                                   #
+#                                                                            #
+# Related work:                                                              #
+# M. Goll, S. Gueron, "Vectorization on ChaCha Stream Cipher", IEEE          #
+#          Proceedings of 11th International Conference on Information       #
+#          Technology: New Generations (ITNG 2014), 612-615 (2014).          #
+# M. Goll, S. Gueron, "Vectorization on Poly1305 Message Authentication Code"#
+#           to be published.                                                 #
+# A. Langley, chacha20poly1305 for the AEAD head                             #
+# https://git.openssl.org/gitweb/?p=openssl.git;a=commit;h=9a8646510b3d0a48e950748f7a2aaa12ed40d5e0  #
+##############################################################################
+
+$flavour = shift;
+$output  = shift;
+if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+
+$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
+( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
+die "can't locate x86_64-xlate.pl";
+
+open OUT,"| \"$^X\" $xlate $flavour $output";
+*STDOUT=*OUT;
+
+if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
+    =~ /GNU assembler version ([2-9]\.[0-9]+)/) {
+  $avx = ($1>=2.19) + ($1>=2.22);
+}
+
+if ($win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
+      `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+  $avx = ($1>=2.09) + ($1>=2.10);
+}
+
+if ($win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
+      `ml64 2>&1` =~ /Version ([0-9]+)\./) {
+  $avx = ($1>=10) + ($1>=11);
+}
+
+if (`$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9])\.([0-9]+)/) {
+  my $ver = $2 + $3/100.0;  # 3.1->3.01, 3.10->3.10
+  $avx = ($ver>=3.0) + ($ver>=3.01);
+}
+
+if ($avx>=2) {{
+
+my ($state_4567, $state_89ab, $state_cdef, $tmp,
+    $v0, $v1, $v2, $v3, $v4, $v5, $v6, $v7,
+    $v8, $v9, $v10, $v11)=map("%ymm$_",(0..15));
+
+sub chacha_qr {
+
+my ($a,$b,$c,$d)=@_;
+
+$code.=<<___;
+
+  vpaddd  $b, $a, $a            # a += b
+  vpxor   $a, $d, $d            # d ^= a
+  vpshufb .rol16(%rip), $d, $d  # d <<<= 16
+
+  vpaddd  $d, $c, $c            # c += d
+  vpxor   $c, $b, $b            # b ^= c
+  vpslld  \$12, $b, $tmp
+  vpsrld  \$20, $b, $b
+  vpxor   $tmp, $b, $b          # b <<<= 12
+
+  vpaddd  $b, $a, $a            # a += b
+  vpxor   $a, $d, $d            # d ^= a
+  vpshufb .rol8(%rip), $d, $d   # d <<<= 8
+
+  vpaddd  $d, $c, $c            # c += d
+  vpxor   $c, $b, $b            # b ^= c
+
+  vpslld  \$7, $b, $tmp
+  vpsrld  \$25, $b, $b
+  vpxor   $tmp, $b, $b          # b <<<= 7
+___
+}
+
+
+$code.=<<___;
+.text
+.align 32
+chacha20_consts:
+.byte 'e','x','p','a','n','d',' ','3','2','-','b','y','t','e',' ','k'
+.byte 'e','x','p','a','n','d',' ','3','2','-','b','y','t','e',' ','k'
+.rol8:
+.byte 3,0,1,2, 7,4,5,6, 11,8,9,10, 15,12,13,14
+.byte 3,0,1,2, 7,4,5,6, 11,8,9,10, 15,12,13,14
+.rol16:
+.byte 2,3,0,1, 6,7,4,5, 10,11,8,9, 14,15,12,13
+.byte 2,3,0,1, 6,7,4,5, 10,11,8,9, 14,15,12,13
+.avx2Init:
+.quad 0,0,1,0
+.avx2Inc:
+.quad 2,0,2,0
+___
+
+{
+
+my $state_cdef_xmm=$state_cdef;
+
+substr($state_cdef_xmm, 1, 1, "x");
+
+my ($out, $in, $in_len, $key_ptr, $nr)
+   =("%rdi", "%rsi", "%rdx", "%rcx", "%r8");
+
+$code.=<<___;
+.globl chacha_20_core_avx2
+.type  chacha_20_core_avx2 ,\@function,2
+.align 64
+chacha_20_core_avx2:
+
+  vzeroupper
+
+  # Init state
+  vbroadcasti128  16*0($key_ptr), $state_4567
+  vbroadcasti128  16*1($key_ptr), $state_89ab
+  vbroadcasti128  16*2($key_ptr), $state_cdef
+  vpaddq    .avx2Init(%rip), $state_cdef, $state_cdef
+
+2:
+  cmp  \$6*64, $in_len
+  jb  2f
+
+  vmovdqa  chacha20_consts(%rip), $v0
+  vmovdqa  chacha20_consts(%rip), $v4
+  vmovdqa  chacha20_consts(%rip), $v8
+
+  vmovdqa  $state_4567, $v1
+  vmovdqa  $state_4567, $v5
+  vmovdqa  $state_4567, $v9
+
+  vmovdqa  $state_89ab, $v2
+  vmovdqa  $state_89ab, $v6
+  vmovdqa  $state_89ab, $v10
+
+  vmovdqa  $state_cdef, $v3
+  vpaddq  .avx2Inc(%rip), $v3, $v7
+  vpaddq  .avx2Inc(%rip), $v7, $v11
+
+  mov  \$10, $nr
+
+  1:
+___
+
+    &chacha_qr( $v0, $v1, $v2, $v3);
+    &chacha_qr( $v4, $v5, $v6, $v7);
+    &chacha_qr( $v8, $v9,$v10,$v11);
+
+$code.=<<___;
+    vpalignr  \$4,  $v1,  $v1,  $v1
+    vpalignr  \$8,  $v2,  $v2,  $v2
+    vpalignr \$12,  $v3,  $v3,  $v3
+    vpalignr  \$4,  $v5,  $v5,  $v5
+    vpalignr  \$8,  $v6,  $v6,  $v6
+    vpalignr \$12,  $v7,  $v7,  $v7
+    vpalignr  \$4,  $v9,  $v9,  $v9
+    vpalignr  \$8, $v10, $v10, $v10
+    vpalignr \$12, $v11, $v11, $v11
+___
+
+    &chacha_qr( $v0, $v1, $v2, $v3);
+    &chacha_qr( $v4, $v5, $v6, $v7);
+    &chacha_qr( $v8, $v9,$v10,$v11);
+
+$code.=<<___;
+    vpalignr \$12,  $v1,  $v1,  $v1
+    vpalignr  \$8,  $v2,  $v2,  $v2
+    vpalignr  \$4,  $v3,  $v3,  $v3
+    vpalignr \$12,  $v5,  $v5,  $v5
+    vpalignr  \$8,  $v6,  $v6,  $v6
+    vpalignr  \$4,  $v7,  $v7,  $v7
+    vpalignr \$12,  $v9,  $v9,  $v9
+    vpalignr  \$8, $v10, $v10, $v10
+    vpalignr  \$4, $v11, $v11, $v11
+
+    dec  $nr
+
+  jnz  1b
+
+  vpaddd  chacha20_consts(%rip), $v0, $v0
+  vpaddd  chacha20_consts(%rip), $v4, $v4
+  vpaddd  chacha20_consts(%rip), $v8, $v8
+
+  vpaddd  $state_4567, $v1, $v1
+  vpaddd  $state_4567, $v5, $v5
+  vpaddd  $state_4567, $v9, $v9
+
+  vpaddd  $state_89ab, $v2, $v2
+  vpaddd  $state_89ab, $v6, $v6
+  vpaddd  $state_89ab, $v10, $v10
+
+  vpaddd  $state_cdef, $v3, $v3
+  vpaddq  .avx2Inc(%rip), $state_cdef, $state_cdef
+  vpaddd  $state_cdef, $v7, $v7
+  vpaddq  .avx2Inc(%rip), $state_cdef, $state_cdef
+  vpaddd  $state_cdef, $v11, $v11
+  vpaddq  .avx2Inc(%rip), $state_cdef, $state_cdef
+
+  vperm2i128  \$0x02, $v0, $v1, $tmp
+  vpxor  32*0($in), $tmp, $tmp
+  vmovdqu  $tmp, 32*0($out)
+  vperm2i128  \$0x02, $v2, $v3, $tmp
+  vpxor  32*1($in), $tmp, $tmp
+  vmovdqu  $tmp, 32*1($out)
+  vperm2i128  \$0x13, $v0, $v1, $tmp
+  vpxor  32*2($in), $tmp, $tmp
+  vmovdqu  $tmp, 32*2($out)
+  vperm2i128  \$0x13, $v2, $v3, $tmp
+  vpxor  32*3($in), $tmp, $tmp
+  vmovdqu  $tmp, 32*3($out)
+
+  vperm2i128  \$0x02, $v4, $v5, $v0
+  vperm2i128  \$0x02, $v6, $v7, $v1
+  vperm2i128  \$0x13, $v4, $v5, $v2
+  vperm2i128  \$0x13, $v6, $v7, $v3
+
+  vpxor  32*4($in), $v0, $v0
+  vpxor  32*5($in), $v1, $v1
+  vpxor  32*6($in), $v2, $v2
+  vpxor  32*7($in), $v3, $v3
+
+  vmovdqu  $v0, 32*4($out)
+  vmovdqu  $v1, 32*5($out)
+  vmovdqu  $v2, 32*6($out)
+  vmovdqu  $v3, 32*7($out)
+
+  vperm2i128  \$0x02, $v8, $v9, $v0
+  vperm2i128  \$0x02, $v10, $v11, $v1
+  vperm2i128  \$0x13, $v8, $v9, $v2
+  vperm2i128  \$0x13, $v10, $v11, $v3
+
+  vpxor  32*8($in), $v0, $v0
+  vpxor  32*9($in), $v1, $v1
+  vpxor  32*10($in), $v2, $v2
+  vpxor  32*11($in), $v3, $v3
+
+  vmovdqu  $v0, 32*8($out)
+  vmovdqu  $v1, 32*9($out)
+  vmovdqu  $v2, 32*10($out)
+  vmovdqu  $v3, 32*11($out)
+
+  lea  64*6($in), $in
+  lea  64*6($out), $out
+  sub  \$64*6, $in_len
+
+  jmp  2b
+
+2:
+  cmp  \$4*64, $in_len
+  jb  2f
+
+  vmovdqa  chacha20_consts(%rip), $v0
+  vmovdqa  chacha20_consts(%rip), $v4
+  vmovdqa  $state_4567, $v1
+  vmovdqa  $state_4567, $v5
+  vmovdqa  $state_89ab, $v2
+  vmovdqa  $state_89ab, $v6
+  vmovdqa  $state_89ab, $v10
+  vmovdqa  $state_cdef, $v3
+  vpaddq   .avx2Inc(%rip), $v3, $v7
+
+  mov  \$10, $nr
+
+  1:
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+    &chacha_qr($v4,$v5,$v6,$v7);
+
+$code.=<<___;
+    vpalignr  \$4, $v1, $v1, $v1
+    vpalignr  \$8, $v2, $v2, $v2
+    vpalignr \$12, $v3, $v3, $v3
+    vpalignr  \$4, $v5, $v5, $v5
+    vpalignr  \$8, $v6, $v6, $v6
+    vpalignr \$12, $v7, $v7, $v7
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+    &chacha_qr($v4,$v5,$v6,$v7);
+
+$code.=<<___;
+    vpalignr \$12, $v1, $v1, $v1
+    vpalignr  \$8, $v2, $v2, $v2
+    vpalignr  \$4, $v3, $v3, $v3
+    vpalignr \$12, $v5, $v5, $v5
+    vpalignr  \$8, $v6, $v6, $v6
+    vpalignr  \$4, $v7, $v7, $v7
+
+    dec  $nr
+
+  jnz  1b
+
+  vpaddd  chacha20_consts(%rip), $v0, $v0
+  vpaddd  chacha20_consts(%rip), $v4, $v4
+
+  vpaddd  $state_4567, $v1, $v1
+  vpaddd  $state_4567, $v5, $v5
+
+  vpaddd  $state_89ab, $v2, $v2
+  vpaddd  $state_89ab, $v6, $v6
+
+  vpaddd  $state_cdef, $v3, $v3
+  vpaddq  .avx2Inc(%rip), $state_cdef, $state_cdef
+  vpaddd  $state_cdef, $v7, $v7
+  vpaddq  .avx2Inc(%rip), $state_cdef, $state_cdef
+
+  vperm2i128  \$0x02, $v0, $v1, $v8
+  vperm2i128  \$0x02, $v2, $v3, $v9
+  vperm2i128  \$0x13, $v0, $v1, $v10
+  vperm2i128  \$0x13, $v2, $v3, $v11
+
+  vpxor  32*0($in), $v8, $v8
+  vpxor  32*1($in), $v9, $v9
+  vpxor  32*2($in), $v10, $v10
+  vpxor  32*3($in), $v11, $v11
+
+  vmovdqu  $v8, 32*0($out)
+  vmovdqu  $v9, 32*1($out)
+  vmovdqu  $v10, 32*2($out)
+  vmovdqu  $v11, 32*3($out)
+
+  vperm2i128  \$0x02, $v4, $v5, $v0
+  vperm2i128  \$0x02, $v6, $v7, $v1
+  vperm2i128  \$0x13, $v4, $v5, $v2
+  vperm2i128  \$0x13, $v6, $v7, $v3
+
+  vpxor  32*4($in), $v0, $v0
+  vpxor  32*5($in), $v1, $v1
+  vpxor  32*6($in), $v2, $v2
+  vpxor  32*7($in), $v3, $v3
+
+  vmovdqu  $v0, 32*4($out)
+  vmovdqu  $v1, 32*5($out)
+  vmovdqu  $v2, 32*6($out)
+  vmovdqu  $v3, 32*7($out)
+
+  lea  64*4($in), $in
+  lea  64*4($out), $out
+  sub  \$64*4, $in_len
+
+  jmp  2b
+2:
+  cmp  \$128, $in_len
+  jb  2f
+
+  vmovdqa  chacha20_consts(%rip), $v0
+  vmovdqa  $state_4567, $v1
+  vmovdqa  $state_89ab, $v2
+  vmovdqa  $state_cdef, $v3
+
+  mov  \$10, $nr
+
+  1:
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+
+$code.=<<___;
+    vpalignr   \$4, $v1, $v1, $v1
+    vpalignr   \$8, $v2, $v2, $v2
+    vpalignr  \$12, $v3, $v3, $v3
+___
+
+    &chacha_qr($v0,$v1,$v2,$v3);
+$code.=<<___;
+    vpalignr  \$12, $v1, $v1, $v1
+    vpalignr   \$8, $v2, $v2, $v2
+    vpalignr   \$4, $v3, $v3, $v3
+
+    dec  $nr
+  jnz  1b
+
+  vpaddd  chacha20_consts(%rip), $v0, $v0
+  vpaddd  $state_4567, $v1, $v1
+  vpaddd  $state_89ab, $v2, $v2
+  vpaddd  $state_cdef, $v3, $v3
+  vpaddq  .avx2Inc(%rip), $state_cdef, $state_cdef
+
+  vperm2i128  \$0x02, $v0, $v1, $v8
+  vperm2i128  \$0x02, $v2, $v3, $v9
+  vperm2i128  \$0x13, $v0, $v1, $v10
+  vperm2i128  \$0x13, $v2, $v3, $v11
+
+  vpxor  32*0($in), $v8, $v8
+  vpxor  32*1($in), $v9, $v9
+  vpxor  32*2($in), $v10, $v10
+  vpxor  32*3($in), $v11, $v11
+
+  vmovdqu  $v8, 32*0($out)
+  vmovdqu  $v9, 32*1($out)
+  vmovdqu  $v10, 32*2($out)
+  vmovdqu  $v11, 32*3($out)
+
+  lea  64*2($in), $in
+  lea  64*2($out), $out
+  sub  \$64*2, $in_len
+  jmp  2b
+
+2:
+  vmovdqu  $state_cdef_xmm, 16*2($key_ptr)
+
+  vzeroupper
+  ret
+.size  chacha_20_core_avx2,.-chacha_20_core_avx2
+___
+}
+}}
+
+
+$code =~ s/\`([^\`]*)\`/eval($1)/gem;
+
+print $code;
+
+close STDOUT;
+
diff --git a/crypto/chacha20poly1305/asm/poly1305_avx.pl b/crypto/chacha20poly1305/asm/poly1305_avx.pl
new file mode 100644
index 0000000..cedeca1
--- /dev/null
+++ b/crypto/chacha20poly1305/asm/poly1305_avx.pl
@@ -0,0 +1,732 @@
+##############################################################################
+#                                                                            #
+# Copyright 2014 Intel Corporation                                           #
+#                                                                            #
+# Licensed under the Apache License, Version 2.0 (the "License");            #
+# you may not use this file except in compliance with the License.           #
+# You may obtain a copy of the License at                                    #
+#                                                                            #
+#    http://www.apache.org/licenses/LICENSE-2.0                              #
+#                                                                            #
+# Unless required by applicable law or agreed to in writing, software        #
+# distributed under the License is distributed on an "AS IS" BASIS,          #
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.   #
+# See the License for the specific language governing permissions and        #
+# limitations under the License.                                             #
+#                                                                            #
+##############################################################################
+#                                                                            #
+#  Developers and authors:                                                   #
+#  Shay Gueron (1, 2), and Vlad Krasnov (1)                                  #
+#  (1) Intel Corporation, Israel Development Center                          #
+#  (2) University of Haifa                                                   #
+#                                                                            #
+##############################################################################
+# state:
+#  0: r[0] || r^2[0]
+# 16: r[1] || r^2[1]
+# 32: r[2] || r^2[2]
+# 48: r[3] || r^2[3]
+# 64: r[4] || r^2[4]
+# 80: r[1]*5 || r^2[1]*5
+# 96: r[2]*5 || r^2[2]*5
+#112: r[3]*5 || r^2[3]*5
+#128: r[4]*5 || r^2[4]*5
+#144: k
+#160: A0
+#164: A1
+#168: A2
+#172: A3
+#176: A4
+#180: END
+
+$flavour = shift;
+$output  = shift;
+if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+
+$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
+( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
+die "can't locate x86_64-xlate.pl";
+
+open OUT,"| \"$^X\" $xlate $flavour $output";
+*STDOUT=*OUT;
+
+if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
+                =~ /GNU assembler version ([2-9]\.[0-9]+)/) {
+        $avx = ($1>=2.19) + ($1>=2.22);
+}
+
+if ($win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
+            `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+        $avx = ($1>=2.09) + ($1>=2.10);
+}
+
+if ($win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
+            `ml64 2>&1` =~ /Version ([0-9]+)\./) {
+        $avx = ($1>=10) + ($1>=11);
+}
+
+if (`$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9])\.([0-9]+)/) {
+        my $ver = $2 + $3/100.0;        # 3.1->3.01, 3.10->3.10
+        $avx = ($ver>=3.0) + ($ver>=3.01);
+}
+
+if ($avx>=1) {{
+
+my ($_r0_, $_r1_, $_r2_, $_r3_, $_r4_,
+    $_r1_x5, $_r2_x5, $_r3_x5, $_r4_x5,
+    $_k_,
+    $_A0_, $_A1_, $_A2_, $_A3_, $_A4_)
+  = (0, 16, 32, 48, 64,
+     80, 96, 112, 128,
+     144,
+     160, 164, 168, 172, 176);
+
+$code.=<<___;
+.text
+.align 32
+.LandMask:
+.quad 0x3FFFFFF, 0x3FFFFFF
+.LsetBit:
+.quad 0x1000000, 0x1000000
+.LrSet:
+.quad 0x0FFFFFFC0FFFFFFF, 0x0FFFFFFC0FFFFFFF
+.quad 0x0FFFFFFC0FFFFFFC, 0x0FFFFFFC0FFFFFFC
+.Lone:
+.quad 1,0
+.Lpoly:
+.quad 0xfffffffffffffffb, 0xffffffffffffffff
+___
+
+
+{
+my ($A0, $A1, $A2, $A3, $A4,
+    $r0, $r1, $r2, $r3, $r4,
+    $T0, $T1, $A5, $A6, $A7, $A8)=map("%xmm$_",(0..15));
+my ($state, $key)
+   =("%rdi", "%rsi");
+
+$code.=<<___;
+################################################################################
+# void poly1305_init_avx(void *state, uint8_t key[32])
+
+.globl poly1305_init_avx
+.type poly1305_init_avx, \@function, 2
+.align 64
+poly1305_init_avx:
+  vzeroupper
+  # load and convert r
+  vmovq   8*0($key), $r0
+  vmovq   8*1($key), $T0
+  vpand   .LrSet(%rip), $r0, $r0
+  vpand   .LrSet+16(%rip), $T0, $T0
+
+  vpsrlq  \$26, $r0, $r1
+  vpand   .LandMask(%rip), $r0, $r0
+  vpsrlq  \$26, $r1, $r2
+  vpand   .LandMask(%rip), $r1, $r1
+  vpsllq  \$12, $T0, $T1
+  vpxor   $T1, $r2, $r2
+  vpsrlq  \$26, $r2, $r3
+  vpsrlq  \$40, $T0, $r4
+  vpand   .LandMask(%rip), $r2, $r2
+  vpand   .LandMask(%rip), $r3, $r3
+
+  # SQR R
+  vpmuludq  $r0, $r0, $A0
+  vpmuludq  $r1, $r0, $A1
+  vpmuludq  $r2, $r0, $A2
+  vpmuludq  $r3, $r0, $A3
+  vpmuludq  $r4, $r0, $A4
+
+  vpsllq    \$1, $A1, $A1
+  vpsllq    \$1, $A2, $A2
+  vpmuludq  $r1, $r1, $T0
+  vpaddq    $T0, $A2, $A2
+  vpmuludq  $r2, $r1, $T0
+  vpaddq    $T0, $A3, $A3
+  vpmuludq  $r3, $r1, $T0
+  vpaddq    $T0, $A4, $A4
+  vpmuludq  $r4, $r1, $A5
+
+  vpsllq    \$1, $A3, $A3
+  vpsllq    \$1, $A4, $A4
+  vpmuludq  $r2, $r2, $T0
+  vpaddq    $T0, $A4, $A4
+  vpmuludq  $r3, $r2, $T0
+  vpaddq    $T0, $A5, $A5
+  vpmuludq  $r4, $r2, $A6
+
+  vpsllq    \$1, $A5, $A5
+  vpsllq    \$1, $A6, $A6
+  vpmuludq  $r3, $r3, $T0
+  vpaddq    $T0, $A6, $A6
+  vpmuludq  $r4, $r3, $A7
+
+  vpsllq    \$1, $A7, $A7
+  vpmuludq  $r4, $r4, $A8
+
+  # Reduce
+  vpsrlq  \$26, $A4, $T0
+  vpand   .LandMask(%rip), $A4, $A4
+  vpaddq  $T0, $A5, $A5
+
+  vpsllq  \$2, $A5, $T0
+  vpaddq  $T0, $A5, $A5
+  vpsllq  \$2, $A6, $T0
+  vpaddq  $T0, $A6, $A6
+  vpsllq  \$2, $A7, $T0
+  vpaddq  $T0, $A7, $A7
+  vpsllq  \$2, $A8, $T0
+  vpaddq  $T0, $A8, $A8
+
+  vpaddq  $A5, $A0, $A0
+  vpaddq  $A6, $A1, $A1
+  vpaddq  $A7, $A2, $A2
+  vpaddq  $A8, $A3, $A3
+
+  vpsrlq  \$26, $A0, $T0
+  vpand   .LandMask(%rip), $A0, $A0
+  vpaddq  $T0, $A1, $A1
+  vpsrlq  \$26, $A1, $T0
+  vpand   .LandMask(%rip), $A1, $A1
+  vpaddq  $T0, $A2, $A2
+  vpsrlq  \$26, $A2, $T0
+  vpand   .LandMask(%rip), $A2, $A2
+  vpaddq  $T0, $A3, $A3
+  vpsrlq  \$26, $A3, $T0
+  vpand   .LandMask(%rip), $A3, $A3
+  vpaddq  $T0, $A4, $A4
+
+  vpunpcklqdq  $r0, $A0, $r0
+  vpunpcklqdq  $r1, $A1, $r1
+  vpunpcklqdq  $r2, $A2, $r2
+  vpunpcklqdq  $r3, $A3, $r3
+  vpunpcklqdq  $r4, $A4, $r4
+
+  vmovdqu $r0, $_r0_($state)
+  vmovdqu $r1, $_r1_($state)
+  vmovdqu $r2, $_r2_($state)
+  vmovdqu $r3, $_r3_($state)
+  vmovdqu $r4, $_r4_($state)
+
+  vpsllq  \$2, $r1, $A1
+  vpsllq  \$2, $r2, $A2
+  vpsllq  \$2, $r3, $A3
+  vpsllq  \$2, $r4, $A4
+
+  vpaddq  $A1, $r1, $A1
+  vpaddq  $A2, $r2, $A2
+  vpaddq  $A3, $r3, $A3
+  vpaddq  $A4, $r4, $A4
+
+  vmovdqu $A1, $_r1_x5($state)
+  vmovdqu $A2, $_r2_x5($state)
+  vmovdqu $A3, $_r3_x5($state)
+  vmovdqu $A4, $_r4_x5($state)
+  # Store k
+  vmovdqu 16*1($key), $T0
+  vmovdqu $T0, $_k_($state)
+  # Init the MAC value
+  vpxor   $T0, $T0, $T0
+  vmovdqu $T0, $_A0_($state)
+  vmovd   $T0, $_A4_($state)
+  vzeroupper
+  ret
+.size poly1305_init_avx,.-poly1305_init_avx
+___
+}
+
+{
+
+my ($A0, $A1, $A2, $A3, $A4,
+    $T0, $T1, $R0, $R1, $R2,
+    $R3, $R4, $AND_MASK)
+   = map("%xmm$_",(0..12));
+
+my ($state, $in, $in_len)
+  =("%rdi", "%rsi", "%rdx");
+
+$code.=<<___;
+
+###############################################################################
+# void* poly1305_update_avx(void* state, void* in, uint64_t in_len)
+.globl poly1305_update_avx
+.type poly1305_update_avx, \@function, 2
+.align 64
+poly1305_update_avx:
+
+  vzeroupper
+  vmovd  $_A0_($state), $A0
+  vmovd  $_A1_($state), $A1
+  vmovd  $_A2_($state), $A2
+  vmovd  $_A3_($state), $A3
+  vmovd  $_A4_($state), $A4
+  vmovdqa  .LandMask(%rip), $AND_MASK
+  # Skip to single block case
+  cmp \$32, $in_len
+  jb  3f
+1:
+    cmp  \$16*4, $in_len
+    jb   1f
+    sub  \$16*2, $in_len
+    # load the next two blocks
+    vmovdqu  16*0($in), $R2
+    vmovdqu  16*1($in), $R3
+    add  \$16*2, $in
+
+    vpunpcklqdq  $R3, $R2, $R0
+    vpunpckhqdq  $R3, $R2, $R1
+
+    vpsrlq  \$26, $R0, $R2
+    vpand   $AND_MASK, $R0, $R0
+    vpaddq  $R0, $A0, $A0
+
+    vpsrlq  \$26, $R2, $R0
+    vpand   $AND_MASK, $R2, $R2
+    vpaddq  $R2, $A1, $A1
+
+    vpsllq  \$12, $R1, $R2
+    vpxor   $R2, $R0, $R0
+    vpand   $AND_MASK, $R0, $R0
+    vpaddq  $R0, $A2, $A2
+
+    vpsrlq  \$26, $R2, $R0
+    vpsrlq  \$40, $R1, $R2
+    vpand   $AND_MASK, $R0, $R0
+    vpxor   .LsetBit(%rip), $R2, $R2
+    vpaddq  $R0, $A3, $A3
+    vpaddq  $R2, $A4, $A4
+
+    # Multiply input by R[0]
+    vbroadcastss  $_r0_($state), $T0
+    vpmuludq  $T0, $A0, $R0
+    vpmuludq  $T0, $A1, $R1
+    vpmuludq  $T0, $A2, $R2
+    vpmuludq  $T0, $A3, $R3
+    vpmuludq  $T0, $A4, $R4
+    # Multiply input by R[1] (and R[1]*5)
+    vbroadcastss  $_r1_x5($state), $T0
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R0, $R0
+    vbroadcastss  $_r1_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R1, $R1
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R2, $R2
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R3, $R3
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R4, $R4
+    # Etc
+    vbroadcastss  $_r2_x5($state), $T0
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R0, $R0
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R1, $R1
+    vbroadcastss  $_r2_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R2, $R2
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R3, $R3
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R4, $R4
+
+    vbroadcastss  $_r3_x5($state), $T0
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R0, $R0
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R1, $R1
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R2, $R2
+    vbroadcastss  $_r3_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R3, $R3
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R4, $R4
+
+    vbroadcastss  $_r4_x5($state), $T0
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R0, $R0
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R1, $R1
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R2, $R2
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R3, $R3
+    vbroadcastss  $_r4_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R4, $R4
+    # Reduce
+    vpsrlq  \$26, $R3, $T0
+    vpaddq  $T0, $R4, $R4
+    vpand   $AND_MASK, $R3, $R3
+
+    vpsrlq  \$26, $R4, $T0
+    vpsllq  \$2, $T0, $T1
+    vpaddq  $T1, $T0, $T0
+    vpaddq  $T0, $R0, $R0
+    vpand   $AND_MASK, $R4, $R4
+
+    vpsrlq  \$26, $R0, $T0
+    vpand   $AND_MASK, $R0, $A0
+    vpaddq  $T0, $R1, $R1
+    vpsrlq  \$26, $R1, $T0
+    vpand   $AND_MASK, $R1, $A1
+    vpaddq  $T0, $R2, $R2
+    vpsrlq  \$26, $R2, $T0
+    vpand   $AND_MASK, $R2, $A2
+    vpaddq  $T0, $R3, $R3
+    vpsrlq  \$26, $R3, $T0
+    vpand   $AND_MASK, $R3, $A3
+    vpaddq  $T0, $R4, $A4
+  jmp 1b
+1:
+  cmp \$16*2, $in_len
+  jb  1f
+  sub \$16*2, $in_len
+  # load the next two blocks
+  vmovdqu 16*0($in), $R2
+  vmovdqu 16*1($in), $R3
+  add \$16*2, $in
+
+  vpunpcklqdq  $R3, $R2, $R0
+  vpunpckhqdq  $R3, $R2, $R1
+
+  vpsrlq  \$26, $R0, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A0, $A0
+
+  vpsrlq  \$26, $R2, $R0
+  vpand   $AND_MASK, $R2, $R2
+  vpaddq  $R2, $A1, $A1
+
+  vpsllq  \$12, $R1, $R2
+  vpxor   $R2, $R0, $R0
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A2, $A2
+
+  vpsrlq  \$26, $R2, $R0
+  vpsrlq  \$40, $R1, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpxor   .LsetBit(%rip), $R2, $R2
+  vpaddq  $R0, $A3, $A3
+  vpaddq  $R2, $A4, $A4
+
+  # Multiply input by R[0]
+  vmovdqu   $_r0_($state), $T0
+  vpmuludq  $T0, $A0, $R0
+  vpmuludq  $T0, $A1, $R1
+  vpmuludq  $T0, $A2, $R2
+  vpmuludq  $T0, $A3, $R3
+  vpmuludq  $T0, $A4, $R4
+  # Multiply input by R[1] (and R[1]*5)
+  vmovdqu   $_r1_x5($state), $T0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R0, $R0
+  vmovdqu   $_r1_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R4, $R4
+  # Etc
+  vmovdqu   $_r2_x5($state), $T0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R1, $R1
+  vmovdqu   $_r2_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovdqu   $_r3_x5($state), $T0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R2, $R2
+  vmovdqu   $_r3_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovdqu   $_r4_x5($state), $T0
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R3, $R3
+  vmovdqu   $_r4_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R4, $R4
+1:
+  vpsrldq  \$8, $R0, $A0
+  vpsrldq  \$8, $R1, $A1
+  vpsrldq  \$8, $R2, $A2
+  vpsrldq  \$8, $R3, $A3
+  vpsrldq  \$8, $R4, $A4
+
+  vpaddq  $R0, $A0, $A0
+  vpaddq  $R1, $A1, $A1
+  vpaddq  $R2, $A2, $A2
+  vpaddq  $R3, $A3, $A3
+  vpaddq  $R4, $A4, $A4
+  # Reduce
+  vpsrlq  \$26, $A3, $T0
+  vpaddq  $T0, $A4, $A4
+  vpand   $AND_MASK, $A3, $A3
+  vpsrlq  \$26, $A4, $T0
+  vpsllq  \$2, $T0, $T1
+  vpaddq  $T1, $T0, $T0
+  vpaddq  $T0, $A0, $A0
+  vpand   $AND_MASK, $A4, $A4
+  vpsrlq  \$26, $A0, $T0
+  vpand   $AND_MASK, $A0, $A0
+  vpaddq  $T0, $A1, $A1
+  vpsrlq  \$26, $A1, $T0
+  vpand   $AND_MASK, $A1, $A1
+  vpaddq  $T0, $A2, $A2
+  vpsrlq  \$26, $A2, $T0
+  vpand   $AND_MASK, $A2, $A2
+  vpaddq  $T0, $A3, $A3
+  vpsrlq  \$26, $A3, $T0
+  vpand   $AND_MASK, $A3, $A3
+  vpaddq  $T0, $A4, $A4
+3:
+  cmp  \$16, $in_len
+  jb  1f
+
+  # load the next block
+  vmovq  8*0($in), $R0
+  vmovq  8*1($in), $R1
+  add    \$16, $in
+  sub    \$16, $in_len
+
+  vpsrlq  \$26, $R0, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A0, $A0
+
+  vpsrlq  \$26, $R2, $R0
+  vpand   $AND_MASK, $R2, $R2
+  vpaddq  $R2, $A1, $A1
+
+  vpsllq  \$12, $R1, $R2
+  vpxor   $R2, $R0, $R0
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A2, $A2
+
+  vpsrlq  \$26, $R2, $R0
+  vpsrlq  \$40, $R1, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpxor   .LsetBit(%rip), $R2, $R2
+  vpaddq  $R0, $A3, $A3
+  vpaddq  $R2, $A4, $A4
+2:
+  # Multiply input by R[0]
+  vmovq     $_r0_+8($state), $T0
+  vpmuludq  $T0, $A0, $R0
+  vpmuludq  $T0, $A1, $R1
+  vpmuludq  $T0, $A2, $R2
+  vpmuludq  $T0, $A3, $R3
+  vpmuludq  $T0, $A4, $R4
+  # Multiply input by R[1] (and R[1]*5)
+  vmovq     $_r1_x5+8($state), $T0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R0, $R0
+  vmovq     $_r1_+8($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R4, $R4
+  # Etc
+  vmovq     $_r2_x5+8($state), $T0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R1, $R1
+  vmovq     $_r2_+8($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovq     $_r3_x5+8($state), $T0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R2, $R2
+  vmovq     $_r3_+8($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovq     $_r4_x5+8($state), $T0
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R3, $R3
+  vmovq     $_r4_+8($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R4, $R4
+  # Reduce
+  vpsrlq  \$26, $R3, $T0
+  vpaddq  $T0, $R4, $R4
+  vpand   $AND_MASK, $R3, $R3
+  vpsrlq  \$26, $R4, $T0
+  vpsllq  \$2, $T0, $T1
+  vpaddq  $T1, $T0, $T0
+  vpaddq  $T0, $R0, $R0
+  vpand   $AND_MASK, $R4, $R4
+  vpsrlq  \$26, $R0, $T0
+  vpand   $AND_MASK, $R0, $A0
+  vpaddq  $T0, $R1, $R1
+  vpsrlq  \$26, $R1, $T0
+  vpand   $AND_MASK, $R1, $A1
+  vpaddq  $T0, $R2, $R2
+  vpsrlq  \$26, $R2, $T0
+  vpand   $AND_MASK, $R2, $A2
+  vpaddq  $T0, $R3, $R3
+  vpsrlq  \$26, $R3, $T0
+  vpand   $AND_MASK, $R3, $A3
+  vpaddq  $T0, $R4, $A4
+
+1:
+  test $in_len, $in_len
+  jz  1f
+
+  vmovdqa  .Lone(%rip), $R0
+3:
+  dec  $in_len
+  vpslldq  \$1, $R0, $R0
+  vpinsrb  \$0, ($in, $in_len), $R0, $R0
+  test $in_len, $in_len
+  jnz  3b
+
+  vpsrldq \$8, $R0, $R1
+  vpsrlq  \$26, $R0, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A0, $A0
+
+  vpsrlq  \$26, $R2, $R0
+  vpand   $AND_MASK, $R2, $R2
+  vpaddq  $R2, $A1, $A1
+
+  vpsllq  \$12, $R1, $R2
+  vpxor   $R2, $R0, $R0
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A2, $A2
+
+  vpsrlq  \$26, $R2, $R0
+  vpsrlq  \$40, $R1, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A3, $A3
+  vpaddq  $R2, $A4, $A4
+  xor  $in_len, $in_len
+  jmp  2b
+1:
+
+  vmovd   $A0, $_A0_($state)
+  vmovd   $A1, $_A1_($state)
+  vmovd   $A2, $_A2_($state)
+  vmovd   $A3, $_A3_($state)
+  vmovd   $A4, $_A4_($state)
+
+  mov $in, %rax
+  vzeroupper
+  ret
+.size poly1305_update_avx,.-poly1305_update_avx
+###############################################################################
+# void poly1305_finish_avx(void* $state, uint64_t mac[2]);
+.type poly1305_finish_avx,\@function, 2
+.globl poly1305_finish_avx
+poly1305_finish_avx:
+___
+my $mac="%rsi";
+$code.=<<___;
+  mov $_A0_($state), %r8d
+  mov $_A1_($state), %eax
+  mov $_A2_($state), %r9d
+  mov $_A3_($state), %ecx
+  mov $_A4_($state), %r10d
+
+  shl \$26, %rax
+  add %rax, %r8
+
+  mov %r9, %rax
+  shl \$52, %rax
+  shr \$12, %r9
+  add %rax, %r8
+  adc \$0, %r9
+
+  mov %r10, %rax
+  shl \$14, %rcx
+  shl \$40, %rax
+  shr \$24, %r10
+
+  add %rcx, %r9
+  add %rax, %r9
+  adc \$0, %r10
+
+  mov %r10, %rax
+  shr \$2, %rax
+  and \$3, %r10
+
+  mov %rax, %rcx
+  shl \$2, %rax
+  add %rcx, %rax
+
+  add %rax, %r8
+  adc \$0, %r9
+  adc \$0, %r10
+
+  mov %r8, %rax
+  mov %r9, %rcx
+  sub \$-5, %rax
+  sbb \$-1, %rcx
+  sbb \$3, %r10
+
+  cmovc %r8, %rax
+  cmovc %r9, %rcx
+  add $_k_($state), %rax
+  adc $_k_+8($state), %rcx
+  mov %rax, ($mac)
+  mov %rcx, 8($mac)
+  ret
+.size poly1305_finish_avx,.-poly1305_finish_avx
+___
+}
+}}
+
+$code =~ s/\`([^\`]*)\`/eval($1)/gem;
+print $code;
+close STDOUT;
+
diff --git a/crypto/chacha20poly1305/asm/poly1305_avx2.pl b/crypto/chacha20poly1305/asm/poly1305_avx2.pl
new file mode 100644
index 0000000..d2dd51f
--- /dev/null
+++ b/crypto/chacha20poly1305/asm/poly1305_avx2.pl
@@ -0,0 +1,984 @@
+##############################################################################
+#                                                                            #
+# Copyright 2014 Intel Corporation                                           #
+#                                                                            #
+# Licensed under the Apache License, Version 2.0 (the "License");            #
+# you may not use this file except in compliance with the License.           #
+# You may obtain a copy of the License at                                    #
+#                                                                            #
+#    http://www.apache.org/licenses/LICENSE-2.0                              #
+#                                                                            #
+# Unless required by applicable law or agreed to in writing, software        #
+# distributed under the License is distributed on an "AS IS" BASIS,          #
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.   #
+# See the License for the specific language governing permissions and        #
+# limitations under the License.                                             #
+#                                                                            #
+##############################################################################
+#                                                                            #
+#  Developers and authors:                                                   #
+#  Shay Gueron (1, 2), and Vlad Krasnov (1)                                  #
+#  (1) Intel Corporation, Israel Development Center                          #
+#  (2) University of Haifa                                                   #
+#                                                                            #
+##############################################################################
+
+$flavour = shift;
+$output  = shift;
+if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+
+$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
+( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
+die "can't locate x86_64-xlate.pl";
+
+open OUT,"| \"$^X\" $xlate $flavour $output";
+*STDOUT=*OUT;
+
+if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
+                =~ /GNU assembler version ([2-9]\.[0-9]+)/) {
+        $avx = ($1>=2.19) + ($1>=2.22);
+}
+
+if ($win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
+            `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+        $avx = ($1>=2.09) + ($1>=2.10);
+}
+
+if ($win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
+            `ml64 2>&1` =~ /Version ([0-9]+)\./) {
+        $avx = ($1>=10) + ($1>=11);
+}
+
+if (`$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9])\.([0-9]+)/) {
+        my $ver = $2 + $3/100.0;        # 3.1->3.01, 3.10->3.10
+        $avx = ($ver>=3.0) + ($ver>=3.01);
+}
+
+if ($avx>=1) {{
+
+my ($_r0_, $_r1_, $_r2_, $_r3_, $_r4_,
+    $_r1_x5, $_r2_x5, $_r3_x5, $_r4_x5,
+    $_k_,
+    $_A0_, $_A1_, $_A2_, $_A3_, $_A4_)
+ = (64, 96, 128, 160, 192,
+    224, 256, 288, 320,
+    40,
+    352, 356, 360, 364, 368);
+
+$code.=<<___;
+.text
+.align 32
+.LandMask:
+.quad 0x3FFFFFF, 0x3FFFFFF, 0x3FFFFFF, 0x3FFFFFF
+.LsetBit:
+.quad 0x1000000, 0x1000000, 0x1000000, 0x1000000
+.LrSet:
+.quad 0xFFFFFFC0FFFFFFF, 0xFFFFFFC0FFFFFFF, 0xFFFFFFC0FFFFFFF, 0xFFFFFFC0FFFFFFF
+.quad 0xFFFFFFC0FFFFFFC, 0xFFFFFFC0FFFFFFC, 0xFFFFFFC0FFFFFFC, 0xFFFFFFC0FFFFFFC
+
+.LpermFix:
+.long 6,7,6,7,6,7,6,7
+.long 4,5,6,7,6,7,6,7
+.long 2,3,6,7,4,5,6,7
+.long 0,1,4,5,2,3,6,7
+___
+
+
+{
+my ($A0, $A1, $A2, $A3, $A4,
+    $r0, $r1, $r2, $r3, $r4,
+    $T0, $T1, $A5, $A6, $A7, $A8)=map("%xmm$_",(0..15));
+
+my ($A0_y, $A1_y, $A2_y, $A3_y, $A4_y,
+    $r0_y, $r1_y, $r2_y, $r3_y, $r4_y)=map("%ymm$_",(0..9));
+
+my ($state, $key)
+   =("%rdi", "%rsi");
+
+$code.=<<___;
+################################################################################
+# void poly1305_init_avx2(void *state)
+
+.type poly1305_init_avx2, \@function, 2
+.align 64
+poly1305_init_avx2:
+  vzeroupper
+
+  # Init the MAC value
+  mov   8*0($state), %r8
+  mov   8*1($state), %r9
+  mov   8*2($state), %r10
+
+  mov   %r8, %rax
+  mov   %r9, %rcx
+
+  sub   \$-5, %r8
+  sbb   \$-1, %r9
+  sbb   \$3, %r10
+
+  cmovc %rax, %r8
+  cmovc %rcx, %r9
+  cmovc 8*2($state), %r10
+
+  mov   %r8, %rax
+  and   \$0x3ffffff, %rax
+  mov   %eax, $_A0_($state)
+
+  shrd  \$26, %r9, %r8
+  shrd  \$26, %r10, %r9
+
+  mov   %r8, %rax
+  and   \$0x3ffffff, %rax
+  mov   %eax, $_A1_($state)
+
+  shrd  \$26, %r9, %r8
+  shr   \$26, %r9
+
+  mov   %r8, %rax
+  and   \$0x3ffffff, %rax
+  mov   %eax, $_A2_($state)
+
+  shrd  \$26, %r9, %r8
+
+  mov   %r8, %rax
+  and   \$0x3ffffff, %rax
+  mov   %eax, $_A3_($state)
+
+  shr   \$26, %r8
+
+  mov   %r8, %rax
+  and   \$0x3ffffff, %rax
+  mov   %eax, $_A4_($state)
+
+  # load and convert r
+  vmovq  8*3($state), $r0
+  vmovq  8*4($state), $T0
+  vpand  .LrSet(%rip), $r0, $r0
+  vpand  .LrSet+32(%rip), $T0, $T0
+
+  vpsrlq \$26, $r0, $r1
+  vpand  .LandMask(%rip), $r0, $r0
+  vpsrlq \$26, $r1, $r2
+  vpand  .LandMask(%rip), $r1, $r1
+  vpsllq \$12, $T0, $T1
+  vpxor  $T1, $r2, $r2
+  vpsrlq \$26, $r2, $r3
+  vpsrlq \$40, $T0, $r4
+  vpand  .LandMask(%rip), $r2, $r2
+  vpand  .LandMask(%rip), $r3, $r3
+  # SQR R
+  vpmuludq  $r0, $r0, $A0
+  vpmuludq  $r1, $r0, $A1
+  vpmuludq  $r2, $r0, $A2
+  vpmuludq  $r3, $r0, $A3
+  vpmuludq  $r4, $r0, $A4
+
+  vpsllq    \$1, $A1, $A1
+  vpsllq    \$1, $A2, $A2
+  vpmuludq  $r1, $r1, $T0
+  vpaddq    $T0, $A2, $A2
+  vpmuludq  $r2, $r1, $T0
+  vpaddq    $T0, $A3, $A3
+  vpmuludq  $r3, $r1, $T0
+  vpaddq    $T0, $A4, $A4
+  vpmuludq  $r4, $r1, $A5
+
+  vpsllq    \$1, $A3, $A3
+  vpsllq    \$1, $A4, $A4
+  vpmuludq  $r2, $r2, $T0
+  vpaddq    $T0, $A4, $A4
+  vpmuludq  $r3, $r2, $T0
+  vpaddq    $T0, $A5, $A5
+  vpmuludq  $r4, $r2, $A6
+
+  vpsllq    \$1, $A5, $A5
+  vpsllq    \$1, $A6, $A6
+  vpmuludq  $r3, $r3, $T0
+  vpaddq    $T0, $A6, $A6
+  vpmuludq  $r4, $r3, $A7
+
+  vpsllq    \$1, $A7, $A7
+  vpmuludq  $r4, $r4, $A8
+
+  # Reduce
+  vpsrlq  \$26, $A4, $T0
+  vpand   .LandMask(%rip), $A4, $A4
+  vpaddq  $T0, $A5, $A5
+
+  vpsllq  \$2, $A5, $T0
+  vpaddq  $T0, $A5, $A5
+  vpsllq  \$2, $A6, $T0
+  vpaddq  $T0, $A6, $A6
+  vpsllq  \$2, $A7, $T0
+  vpaddq  $T0, $A7, $A7
+  vpsllq  \$2, $A8, $T0
+  vpaddq  $T0, $A8, $A8
+
+  vpaddq  $A5, $A0, $A0
+  vpaddq  $A6, $A1, $A1
+  vpaddq  $A7, $A2, $A2
+  vpaddq  $A8, $A3, $A3
+
+  vpsrlq  \$26, $A0, $T0
+  vpand   .LandMask(%rip), $A0, $A0
+  vpaddq  $T0, $A1, $A1
+  vpsrlq  \$26, $A1, $T0
+  vpand   .LandMask(%rip), $A1, $A1
+  vpaddq  $T0, $A2, $A2
+  vpsrlq  \$26, $A2, $T0
+  vpand   .LandMask(%rip), $A2, $A2
+  vpaddq  $T0, $A3, $A3
+  vpsrlq  \$26, $A3, $T0
+  vpand   .LandMask(%rip), $A3, $A3
+  vpaddq  $T0, $A4, $A4
+
+  vpunpcklqdq  $r0, $A0, $r0
+  vpunpcklqdq  $r1, $A1, $r1
+  vpunpcklqdq  $r2, $A2, $r2
+  vpunpcklqdq  $r3, $A3, $r3
+  vpunpcklqdq  $r4, $A4, $r4
+
+  vmovdqu $r0, $_r0_+16($state)
+  vmovdqu $r1, $_r1_+16($state)
+  vmovdqu $r2, $_r2_+16($state)
+  vmovdqu $r3, $_r3_+16($state)
+  vmovdqu $r4, $_r4_+16($state)
+
+  vpsllq  \$2, $r1, $A1
+  vpsllq  \$2, $r2, $A2
+  vpsllq  \$2, $r3, $A3
+  vpsllq  \$2, $r4, $A4
+
+  vpaddq  $A1, $r1, $A1
+  vpaddq  $A2, $r2, $A2
+  vpaddq  $A3, $r3, $A3
+  vpaddq  $A4, $r4, $A4
+
+  vmovdqu $A1, $_r1_x5+16($state)
+  vmovdqu $A2, $_r2_x5+16($state)
+  vmovdqu $A3, $_r3_x5+16($state)
+  vmovdqu $A4, $_r4_x5+16($state)
+
+  # Compute r^3 and r^4
+  vpshufd \$0x44, $r0, $A0
+  vpshufd \$0x44, $r1, $A1
+  vpshufd \$0x44, $r2, $A2
+  vpshufd \$0x44, $r3, $A3
+  vpshufd \$0x44, $r4, $A4
+
+  # Multiply input by R[0]
+  vmovdqu  $_r0_+16($state), $T0
+  vpmuludq $T0, $A0, $r0
+  vpmuludq $T0, $A1, $r1
+  vpmuludq $T0, $A2, $r2
+  vpmuludq $T0, $A3, $r3
+  vpmuludq $T0, $A4, $r4
+  # Multiply input by R[1] (and R[1]*5)
+  vmovdqu  $_r1_x5+16($state), $T0
+  vpmuludq $T0, $A4, $T1
+  vpaddq   $T1, $r0, $r0
+  vmovdqu  $_r1_+16($state), $T0
+  vpmuludq $T0, $A0, $T1
+  vpaddq   $T1, $r1, $r1
+  vpmuludq $T0, $A1, $T1
+  vpaddq   $T1, $r2, $r2
+  vpmuludq $T0, $A2, $T1
+  vpaddq   $T1, $r3, $r3
+  vpmuludq $T0, $A3, $T1
+  vpaddq   $T1, $r4, $r4
+  # Etc
+  vmovdqu  $_r2_x5+16($state), $T0
+  vpmuludq $T0, $A3, $T1
+  vpaddq   $T1, $r0, $r0
+  vpmuludq $T0, $A4, $T1
+  vpaddq   $T1, $r1, $r1
+  vmovdqu  $_r2_+16($state), $T0
+  vpmuludq $T0, $A0, $T1
+  vpaddq   $T1, $r2, $r2
+  vpmuludq $T0, $A1, $T1
+  vpaddq   $T1, $r3, $r3
+  vpmuludq $T0, $A2, $T1
+  vpaddq   $T1, $r4, $r4
+
+  vmovdqu  $_r3_x5+16($state), $T0
+  vpmuludq $T0, $A2, $T1
+  vpaddq   $T1, $r0, $r0
+  vpmuludq $T0, $A3, $T1
+  vpaddq   $T1, $r1, $r1
+  vpmuludq $T0, $A4, $T1
+  vpaddq   $T1, $r2, $r2
+  vmovdqu  $_r3_+16($state), $T0
+  vpmuludq $T0, $A0, $T1
+  vpaddq   $T1, $r3, $r3
+  vpmuludq $T0, $A1, $T1
+  vpaddq   $T1, $r4, $r4
+
+  vmovdqu  $_r4_x5+16($state), $T0
+  vpmuludq $T0, $A1, $T1
+  vpaddq   $T1, $r0, $r0
+  vpmuludq $T0, $A2, $T1
+  vpaddq   $T1, $r1, $r1
+  vpmuludq $T0, $A3, $T1
+  vpaddq   $T1, $r2, $r2
+  vpmuludq $T0, $A4, $T1
+  vpaddq   $T1, $r3, $r3
+  vmovdqu  $_r4_+16($state), $T0
+  vpmuludq $T0, $A0, $T1
+  vpaddq   $T1, $r4, $r4
+  # Reduce
+  vpsrlq  \$26, $r3, $T0
+  vpaddq  $T0, $r4, $r4
+  vpand   .LandMask(%rip), $r3, $r3
+  vpsrlq  \$26, $r4, $T0
+  vpsllq  \$2, $T0, $T1
+  vpaddq  $T1, $T0, $T0
+  vpaddq  $T0, $r0, $r0
+  vpand   .LandMask(%rip), $r4, $r4
+  vpsrlq  \$26, $r0, $T0
+  vpand   .LandMask(%rip), $r0, $r0
+  vpaddq  $T0, $r1, $r1
+  vpsrlq  \$26, $r1, $T0
+  vpand   .LandMask(%rip), $r1, $r1
+  vpaddq  $T0, $r2, $r2
+  vpsrlq  \$26, $r2, $T0
+  vpand   .LandMask(%rip), $r2, $r2
+  vpaddq  $T0, $r3, $r3
+  vpsrlq  \$26, $r3, $T0
+  vpand   .LandMask(%rip), $r3, $r3
+  vpaddq  $T0, $r4, $r4
+
+  vmovdqu $r0, $_r0_($state)
+  vmovdqu $r1, $_r1_($state)
+  vmovdqu $r2, $_r2_($state)
+  vmovdqu $r3, $_r3_($state)
+  vmovdqu $r4, $_r4_($state)
+
+  vpsllq  \$2, $r1, $A1
+  vpsllq  \$2, $r2, $A2
+  vpsllq  \$2, $r3, $A3
+  vpsllq  \$2, $r4, $A4
+
+  vpaddq  $A1, $r1, $A1
+  vpaddq  $A2, $r2, $A2
+  vpaddq  $A3, $r3, $A3
+  vpaddq  $A4, $r4, $A4
+
+  vmovdqu $A1, $_r1_x5($state)
+  vmovdqu $A2, $_r2_x5($state)
+  vmovdqu $A3, $_r3_x5($state)
+  vmovdqu $A4, $_r4_x5($state)
+
+  movq  \$1, 8*7($state)
+
+  ret
+.size poly1305_init_avx2,.-poly1305_init_avx2
+___
+}
+
+{
+
+my ($A0, $A1, $A2, $A3, $A4,
+    $T0, $T1, $R0, $R1, $R2,
+    $R3, $R4, $AND_MASK, $PERM_MASK, $SET_MASK)
+  =map("%ymm$_",(0..14));
+
+my ($A0_x, $A1_x, $A2_x, $A3_x, $A4_x,
+    $T0_x, $T1_x, $R0_x, $R1_x, $R2_x,
+    $R3_x, $R4_x, $AND_MASK_x, $PERM_MASK_x, $SET_MASK_x)
+  =map("%xmm$_",(0..14));
+
+my ($state, $in, $in_len, $hlp, $rsp_save)
+  =("%rdi", "%rsi", "%rdx", "%rcx", "%rax");
+
+$code.=<<___;
+
+###############################################################################
+# void poly1305_update_avx2(void* $state, void* in, uint64_t in_len)
+.globl poly1305_update_avx2
+.type poly1305_update_avx2, \@function, 2
+.align 64
+poly1305_update_avx2:
+
+
+  test  $in_len, $in_len
+  jz    6f
+
+  cmpq  \$0, 8*7($state)
+  jne   1f  # This means we already started avx2, and must finish
+
+  # If we are here we need to check in_len is longer than the min for avx2
+
+  cmp   \$512, $in_len
+  jge   2f
+
+  jmp   poly1305_update_x64
+
+2:
+  call poly1305_init_avx2
+
+1:
+  vmovd  $_A0_($state), $A0_x
+  vmovd  $_A1_($state), $A1_x
+  vmovd  $_A2_($state), $A2_x
+  vmovd  $_A3_($state), $A3_x
+  vmovd  $_A4_($state), $A4_x
+
+  vmovdqa  .LandMask(%rip), $AND_MASK
+1:
+    cmp \$32*4, $in_len
+    jb  1f
+    sub \$32*2, $in_len
+
+    # load the next four blocks
+    vmovdqu 32*0($in), $R2
+    vmovdqu 32*1($in), $R3
+    add \$32*2, $in
+
+    vpunpcklqdq  $R3, $R2, $R0
+    vpunpckhqdq  $R3, $R2, $R1
+
+    vpermq  \$0xD8, $R0, $R0  # it is possible to rearrange the precomputations, and save this shuffle
+    vpermq  \$0xD8, $R1, $R1
+
+    vpsrlq  \$26, $R0, $R2
+    vpand   $AND_MASK, $R0, $R0
+    vpaddq  $R0, $A0, $A0
+
+    vpsrlq  \$26, $R2, $R0
+    vpand   $AND_MASK, $R2, $R2
+    vpaddq  $R2, $A1, $A1
+
+    vpsllq  \$12, $R1, $R2
+    vpxor   $R2, $R0, $R0
+    vpand   $AND_MASK, $R0, $R0
+    vpaddq  $R0, $A2, $A2
+
+    vpsrlq  \$26, $R2, $R0
+    vpsrlq  \$40, $R1, $R2
+    vpand   $AND_MASK, $R0, $R0
+    vpxor   .LsetBit(%rip), $R2, $R2
+    vpaddq  $R0, $A3, $A3
+    vpaddq  $R2, $A4, $A4
+
+    # Multiply input by R[0]
+    vpbroadcastq  $_r0_($state), $T0
+    vpmuludq  $T0, $A0, $R0
+    vpmuludq  $T0, $A1, $R1
+    vpmuludq  $T0, $A2, $R2
+    vpmuludq  $T0, $A3, $R3
+    vpmuludq  $T0, $A4, $R4
+    # Multiply input by R[1] (and R[1]*5)
+    vpbroadcastq  $_r1_x5($state), $T0
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R0, $R0
+    vpbroadcastq  $_r1_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R1, $R1
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R2, $R2
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R3, $R3
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R4, $R4
+    # Etc
+    vpbroadcastq  $_r2_x5($state), $T0
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R0, $R0
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R1, $R1
+    vpbroadcastq  $_r2_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R2, $R2
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R3, $R3
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R4, $R4
+
+    vpbroadcastq  $_r3_x5($state), $T0
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R0, $R0
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R1, $R1
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R2, $R2
+    vpbroadcastq  $_r3_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R3, $R3
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R4, $R4
+
+    vpbroadcastq  $_r4_x5($state), $T0
+    vpmuludq  $T0, $A1, $T1
+    vpaddq    $T1, $R0, $R0
+    vpmuludq  $T0, $A2, $T1
+    vpaddq    $T1, $R1, $R1
+    vpmuludq  $T0, $A3, $T1
+    vpaddq    $T1, $R2, $R2
+    vpmuludq  $T0, $A4, $T1
+    vpaddq    $T1, $R3, $R3
+    vpbroadcastq  $_r4_($state), $T0
+    vpmuludq  $T0, $A0, $T1
+    vpaddq    $T1, $R4, $R4
+    # Reduce
+    vpsrlq  \$26, $R3, $T0
+    vpaddq  $T0, $R4, $R4
+    vpand   $AND_MASK, $R3, $R3
+
+    vpsrlq  \$26, $R4, $T0
+    vpsllq  \$2, $T0, $T1
+    vpaddq  $T1, $T0, $T0
+    vpaddq  $T0, $R0, $R0
+    vpand   $AND_MASK, $R4, $R4
+
+    vpsrlq  \$26, $R0, $T0
+    vpand   $AND_MASK, $R0, $A0
+    vpaddq  $T0, $R1, $R1
+    vpsrlq  \$26, $R1, $T0
+    vpand   $AND_MASK, $R1, $A1
+    vpaddq  $T0, $R2, $R2
+    vpsrlq  \$26, $R2, $T0
+    vpand   $AND_MASK, $R2, $A2
+    vpaddq  $T0, $R3, $R3
+    vpsrlq  \$26, $R3, $T0
+    vpand   $AND_MASK, $R3, $A3
+    vpaddq  $T0, $R4, $A4
+  jmp 1b
+1:
+
+  cmp  \$32*2, $in_len
+  jb  1f
+  sub  \$32*2, $in_len
+  # load the next four blocks
+  vmovdqu  32*0($in), $R2
+  vmovdqu  32*1($in), $R3
+  add  \$32*2, $in
+
+  vpunpcklqdq  $R3, $R2, $R0
+  vpunpckhqdq  $R3, $R2, $R1
+
+  vpermq  \$0xD8, $R0, $R0
+  vpermq  \$0xD8, $R1, $R1
+
+  vpsrlq  \$26, $R0, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A0, $A0
+
+  vpsrlq  \$26, $R2, $R0
+  vpand   $AND_MASK, $R2, $R2
+  vpaddq  $R2, $A1, $A1
+
+  vpsllq  \$12, $R1, $R2
+  vpxor   $R2, $R0, $R0
+  vpand   $AND_MASK, $R0, $R0
+  vpaddq  $R0, $A2, $A2
+
+  vpsrlq  \$26, $R2, $R0
+  vpsrlq  \$40, $R1, $R2
+  vpand   $AND_MASK, $R0, $R0
+  vpxor   .LsetBit(%rip), $R2, $R2
+  vpaddq  $R0, $A3, $A3
+  vpaddq  $R2, $A4, $A4
+
+  # Multiply input by R[0]
+  vmovdqu   $_r0_($state), $T0
+  vpmuludq  $T0, $A0, $R0
+  vpmuludq  $T0, $A1, $R1
+  vpmuludq  $T0, $A2, $R2
+  vpmuludq  $T0, $A3, $R3
+  vpmuludq  $T0, $A4, $R4
+  # Multiply input by R[1] (and R[1]*5)
+  vmovdqu   $_r1_x5($state), $T0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R0, $R0
+  vmovdqu   $_r1_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R4, $R4
+  # Etc
+  vmovdqu   $_r2_x5($state), $T0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R1, $R1
+  vmovdqu   $_r2_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovdqu   $_r3_x5($state), $T0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R2, $R2
+  vmovdqu   $_r3_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovdqu   $_r4_x5($state), $T0
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R3, $R3
+  vmovdqu   $_r4_($state), $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R4, $R4
+  # Reduce
+  vpsrlq  \$26, $R3, $T0
+  vpaddq  $T0, $R4, $R4
+  vpand   $AND_MASK, $R3, $R3
+  vpsrlq  \$26, $R4, $T0
+  vpsllq  \$2, $T0, $T1
+  vpaddq  $T1, $T0, $T0
+  vpaddq  $T0, $R0, $R0
+  vpand   $AND_MASK, $R4, $R4
+  vpsrlq  \$26, $R0, $T0
+  vpand   $AND_MASK, $R0, $A0
+  vpaddq  $T0, $R1, $R1
+  vpsrlq  \$26, $R1, $T0
+  vpand   $AND_MASK, $R1, $A1
+  vpaddq  $T0, $R2, $R2
+  vpsrlq  \$26, $R2, $T0
+  vpand   $AND_MASK, $R2, $A2
+  vpaddq  $T0, $R3, $R3
+  vpsrlq  \$26, $R3, $T0
+  vpand   $AND_MASK, $R3, $A3
+  vpaddq  $T0, $R4, $A4
+
+  vpsrldq \$8, $A0, $R0
+  vpsrldq \$8, $A1, $R1
+  vpsrldq \$8, $A2, $R2
+  vpsrldq \$8, $A3, $R3
+  vpsrldq \$8, $A4, $R4
+
+  vpaddq  $R0, $A0, $A0
+  vpaddq  $R1, $A1, $A1
+  vpaddq  $R2, $A2, $A2
+  vpaddq  $R3, $A3, $A3
+  vpaddq  $R4, $A4, $A4
+
+  vpermq  \$0xAA, $A0, $R0
+  vpermq  \$0xAA, $A1, $R1
+  vpermq  \$0xAA, $A2, $R2
+  vpermq  \$0xAA, $A3, $R3
+  vpermq  \$0xAA, $A4, $R4
+
+  vpaddq  $R0, $A0, $A0
+  vpaddq  $R1, $A1, $A1
+  vpaddq  $R2, $A2, $A2
+  vpaddq  $R3, $A3, $A3
+  vpaddq  $R4, $A4, $A4
+1:
+  test  $in_len, $in_len
+  jz  5f
+  # In case 1,2 or 3 blocks remain, we want to multiply them correctly
+  vmovq  $A0_x, $A0_x
+  vmovq  $A1_x, $A1_x
+  vmovq  $A2_x, $A2_x
+  vmovq  $A3_x, $A3_x
+  vmovq  $A4_x, $A4_x
+
+  mov  .LsetBit(%rip), $hlp
+  mov  %rsp, $rsp_save
+  test \$15, $in_len
+  jz   1f
+  xor  $hlp, $hlp
+  sub  \$64, %rsp
+  vpxor   $R0, $R0, $R0
+  vmovdqu $R0, (%rsp)
+  vmovdqu $R0, 32(%rsp)
+3:
+  movb  ($in, $hlp), %r8b
+  movb  %r8b, (%rsp, $hlp)
+  inc  $hlp
+  cmp  $hlp, $in_len
+  jne  3b
+
+  movb \$1, (%rsp, $hlp)
+  xor  $hlp, $hlp
+  mov  %rsp, $in
+
+1:
+
+  cmp  \$16, $in_len
+  ja  2f
+  vmovq  8*0($in), $R0_x
+  vmovq  8*1($in), $R1_x
+  vmovq  $hlp, $SET_MASK_x
+  vmovdqa  .LpermFix(%rip), $PERM_MASK
+  jmp  1f
+2:
+  cmp  \$32, $in_len
+  ja  2f
+  vmovdqu 16*0($in), $R2_x
+  vmovdqu 16*1($in), $R3_x
+  vmovq   .LsetBit(%rip), $SET_MASK_x
+  vpinsrq  \$1, $hlp, $SET_MASK_x, $SET_MASK_x
+  vmovdqa .LpermFix+32(%rip), $PERM_MASK
+
+  vpunpcklqdq  $R3, $R2, $R0
+  vpunpckhqdq  $R3, $R2, $R1
+  jmp  1f
+2:
+  cmp  \$48, $in_len
+  ja  2f
+  vmovdqu  32*0($in), $R2
+  vmovdqu  32*1($in), $R3_x
+  vmovq  .LsetBit(%rip), $SET_MASK_x
+  vpinsrq \$1, $hlp, $SET_MASK_x, $SET_MASK_x
+  vpermq  \$0xc4, $SET_MASK, $SET_MASK
+  vmovdqa  .LpermFix+64(%rip), $PERM_MASK
+
+  vpunpcklqdq  $R3, $R2, $R0
+  vpunpckhqdq  $R3, $R2, $R1
+  jmp  1f
+2:
+  vmovdqu 32*0($in), $R2
+  vmovdqu 32*1($in), $R3
+  vmovq   .LsetBit(%rip), $SET_MASK_x
+  vpinsrq \$1, $hlp, $SET_MASK_x, $SET_MASK_x
+  vpermq  \$0x40, $SET_MASK, $SET_MASK
+  vmovdqa .LpermFix+96(%rip), $PERM_MASK
+
+  vpunpcklqdq     $R3, $R2, $R0
+  vpunpckhqdq     $R3, $R2, $R1
+
+1:
+  mov  $rsp_save, %rsp
+
+  vpsrlq \$26, $R0, $R2
+  vpand  $AND_MASK, $R0, $R0
+  vpaddq $R0, $A0, $A0
+
+  vpsrlq \$26, $R2, $R0
+  vpand  $AND_MASK, $R2, $R2
+  vpaddq $R2, $A1, $A1
+
+  vpsllq \$12, $R1, $R2
+  vpxor  $R2, $R0, $R0
+  vpand  $AND_MASK, $R0, $R0
+  vpaddq $R0, $A2, $A2
+
+  vpsrlq \$26, $R2, $R0
+  vpsrlq \$40, $R1, $R2
+  vpand  $AND_MASK, $R0, $R0
+  vpxor  $SET_MASK, $R2, $R2
+  vpaddq $R0, $A3, $A3
+  vpaddq $R2, $A4, $A4
+
+  # Multiply input by R[0]
+  vmovdqu   $_r0_($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A0, $R0
+  vpmuludq  $T0, $A1, $R1
+  vpmuludq  $T0, $A2, $R2
+  vpmuludq  $T0, $A3, $R3
+  vpmuludq  $T0, $A4, $R4
+  # Multiply input by R[1] (and R[1]*5)
+  vmovdqu   $_r1_x5($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R0, $R0
+  vmovdqu   $_r1_($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R4, $R4
+  # Etc
+  vmovdqu   $_r2_x5($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R1, $R1
+  vmovdqu   $_r2_($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovdqu   $_r3_x5($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R2, $R2
+  vmovdqu   $_r3_($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R3, $R3
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R4, $R4
+
+  vmovdqu   $_r4_x5($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A1, $T1
+  vpaddq    $T1, $R0, $R0
+  vpmuludq  $T0, $A2, $T1
+  vpaddq    $T1, $R1, $R1
+  vpmuludq  $T0, $A3, $T1
+  vpaddq    $T1, $R2, $R2
+  vpmuludq  $T0, $A4, $T1
+  vpaddq    $T1, $R3, $R3
+  vmovdqu   $_r4_($state), $T0
+  vpermd    $T0, $PERM_MASK, $T0
+  vpmuludq  $T0, $A0, $T1
+  vpaddq    $T1, $R4, $R4
+  # Reduce
+  vpsrlq  \$26, $R3, $T0
+  vpaddq  $T0, $R4, $R4
+  vpand   $AND_MASK, $R3, $R3
+  vpsrlq  \$26, $R4, $T0
+  vpsllq  \$2, $T0, $T1
+  vpaddq  $T1, $T0, $T0
+  vpaddq  $T0, $R0, $R0
+  vpand   $AND_MASK, $R4, $R4
+  vpsrlq  \$26, $R0, $T0
+  vpand   $AND_MASK, $R0, $A0
+  vpaddq  $T0, $R1, $R1
+  vpsrlq  \$26, $R1, $T0
+  vpand   $AND_MASK, $R1, $A1
+  vpaddq  $T0, $R2, $R2
+  vpsrlq  \$26, $R2, $T0
+  vpand   $AND_MASK, $R2, $A2
+  vpaddq  $T0, $R3, $R3
+  vpsrlq  \$26, $R3, $T0
+  vpand   $AND_MASK, $R3, $A3
+  vpaddq  $T0, $R4, $A4
+
+  vpsrldq \$8, $A0, $R0
+  vpsrldq \$8, $A1, $R1
+  vpsrldq \$8, $A2, $R2
+  vpsrldq \$8, $A3, $R3
+  vpsrldq \$8, $A4, $R4
+
+  vpaddq  $R0, $A0, $A0
+  vpaddq  $R1, $A1, $A1
+  vpaddq  $R2, $A2, $A2
+  vpaddq  $R3, $A3, $A3
+  vpaddq  $R4, $A4, $A4
+
+  vpermq  \$0xAA, $A0, $R0
+  vpermq  \$0xAA, $A1, $R1
+  vpermq  \$0xAA, $A2, $R2
+  vpermq  \$0xAA, $A3, $R3
+  vpermq  \$0xAA, $A4, $R4
+
+  vpaddq  $R0, $A0, $A0
+  vpaddq  $R1, $A1, $A1
+  vpaddq  $R2, $A2, $A2
+  vpaddq  $R3, $A3, $A3
+  vpaddq  $R4, $A4, $A4
+
+5:
+  vmovd  $A0_x, $_A0_($state)
+  vmovd  $A1_x, $_A1_($state)
+  vmovd  $A2_x, $_A2_($state)
+  vmovd  $A3_x, $_A3_($state)
+  vmovd  $A4_x, $_A4_($state)
+6:
+  ret
+.size poly1305_update_avx2,.-poly1305_update_avx2
+###############################################################################
+# void poly1305_finish_avx2(void* $state, uint8_t mac[16]);
+.type poly1305_finish_avx2,\@function,2
+.globl poly1305_finish_avx2
+poly1305_finish_avx2:
+___
+my $mac="%rsi";
+my ($A0, $A1, $A2, $A3, $A4, $T0, $T1)
+  =map("%xmm$_",(0..6));
+
+$code.=<<___;
+  cmpq  \$0, 8*7($state)
+  jne 1f
+  jmp poly1305_finish_x64
+
+1:
+  mov  $_A0_($state), %r8d
+  mov  $_A1_($state), %eax
+  mov  $_A2_($state), %r9d
+  mov  $_A3_($state), %ecx
+  mov  $_A4_($state), %r10d
+
+  shl \$26, %rax
+  add %rax, %r8
+
+  mov %r9, %rax
+  shl \$52, %rax
+  shr \$12, %r9
+  add %rax, %r8
+  adc \$0, %r9
+
+  mov %r10, %rax
+  shl \$14, %rcx
+  shl \$40, %rax
+  shr \$24, %r10
+
+  add %rcx, %r9
+  add %rax, %r9
+  adc \$0, %r10
+
+  mov %r10, %rax
+  shr \$2, %rax
+  and \$3, %r10
+
+  mov %rax, %rcx
+  shl \$2, %rax
+  add %rcx, %rax
+
+  add %rax, %r8
+  adc \$0, %r9
+  adc \$0, %r10
+
+  mov %r8, %rax
+  mov %r9, %rcx
+  sub \$-5, %rax
+  sbb \$-1, %rcx
+  sbb \$3, %r10
+
+  cmovc %r8, %rax
+  cmovc %r9, %rcx
+  add  $_k_($state), %rax
+  adc  $_k_+8($state), %rcx
+  mov %rax, ($mac)
+  mov %rcx, 8($mac)
+
+  ret
+.size poly1305_finish_avx2,.-poly1305_finish_avx2
+___
+}
+}}
+
+$code =~ s/\`([^\`]*)\`/eval(\$1)/gem;
+print $code;
+close STDOUT;
+
diff --git a/crypto/chacha20poly1305/asm/poly1305_x64.pl b/crypto/chacha20poly1305/asm/poly1305_x64.pl
new file mode 100644
index 0000000..31c4c47
--- /dev/null
+++ b/crypto/chacha20poly1305/asm/poly1305_x64.pl
@@ -0,0 +1,281 @@
+
+##############################################################################
+#                                                                            #
+# Copyright 2016 CloudFlare LTD                                              #
+#                                                                            #
+# Licensed under the Apache License, Version 2.0 (the "License");            #
+# you may not use this file except in compliance with the License.           #
+# You may obtain a copy of the License at                                    #
+#                                                                            #
+#    http://www.apache.org/licenses/LICENSE-2.0                              #
+#                                                                            #
+# Unless required by applicable law or agreed to in writing, software        #
+# distributed under the License is distributed on an "AS IS" BASIS,          #
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.   #
+# See the License for the specific language governing permissions and        #
+# limitations under the License.                                             #
+#                                                                            #
+##############################################################################
+#                                                                            #
+# Author:  Vlad Krasnov                                                      #
+#                                                                            #
+##############################################################################
+
+$flavour = shift;
+$output  = shift;
+if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
+
+$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
+( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
+die "can't locate x86_64-xlate.pl";
+
+open OUT,"| \"$^X\" $xlate $flavour $output";
+*STDOUT=*OUT;
+
+if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
+                =~ /GNU assembler version ([2-9]\.[0-9]+)/) {
+        $avx = ($1>=2.19) + ($1>=2.22);
+}
+
+if ($win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
+            `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+        $avx = ($1>=2.09) + ($1>=2.10);
+}
+
+if ($win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
+            `ml64 2>&1` =~ /Version ([0-9]+)\./) {
+        $avx = ($1>=10) + ($1>=11);
+}
+
+if (`$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9])\.([0-9]+)/) {
+        my $ver = $2 + $3/100.0;        # 3.1->3.01, 3.10->3.10
+        $avx = ($ver>=3.0) + ($ver>=3.01);
+}
+
+
+{
+{
+
+my ($state, $key)
+   =("%rdi", "%rsi");
+
+$code.=<<___;
+
+.LrSet:
+.align 16
+.quad 0x0FFFFFFC0FFFFFFF, 0x0FFFFFFC0FFFFFFC
+###############################################################################
+# void poly1305_init_x64(void *state, uint8_t key[32])
+
+.globl poly1305_init_x64
+.type poly1305_init_x64, \@function, 2
+.align 64
+poly1305_init_x64:
+
+  xor %rax, %rax
+  mov %rax, 8*0($state)
+  mov %rax, 8*1($state)
+  mov %rax, 8*2($state)
+
+  movdqu 16*0($key), %xmm0
+  movdqu 16*1($key), %xmm1
+  pand .LrSet(%rip), %xmm0
+
+  movdqu %xmm0, 8*3($state)
+  movdqu %xmm1, 8*3+16($state)
+  movq  \$0, 8*7($state)
+
+  ret
+.size poly1305_init_x64,.-poly1305_init_x64
+___
+}
+
+{
+
+my ($state, $inp)
+   =("%rdi", "%rsi");
+
+my ($acc0, $acc1, $acc2, $inl, $t0, $t1, $t2, $t3, $r0)
+   =("%rcx", "%r8", "%r9", "%r10", "%r11", "%r12", "%r13", "%r14", "%r15");
+
+my ($r1)
+   =("8*4($state)");
+
+$code.=<<___;
+###############################################################################
+# void* poly1305_update_x64(void* state, void* in, uint64_t in_len)
+.globl poly1305_update_x64
+.type poly1305_update_x64, \@function, 2
+.align 64
+poly1305_update_x64:
+
+  push  %r11
+  push  %r12
+  push  %r13
+  push  %r14
+  push  %r15
+
+  mov %rdx, $inl
+
+  mov 8*0($state), $acc0
+  mov 8*1($state), $acc1
+  mov 8*2($state), $acc2
+  mov 8*3($state), $r0
+
+  cmp   \$16, $inl
+  jb    2f
+  jmp   1f
+
+.align 64
+1:
+############################
+  add   8*0($inp), $acc0
+  adc   8*1($inp), $acc1
+  lea   16($inp), $inp
+  adc   \$1, $acc2
+
+5:
+  mov   $r0, %rax
+  mulq  $acc0
+  mov   %rax, $t0
+  mov   %rdx, $t1
+
+  mov   $r0, %rax
+  mulq  $acc1
+  add   %rax, $t1
+  adc   \$0, %rdx
+
+  mov   $r0, $t2
+  imul  $acc2, $t2
+  add   %rdx, $t2
+############################
+  mov   $r1, %rax
+  mulq  $acc0
+  add   %rax, $t1
+  adc   \$0, %rdx
+  mov   %rdx, $acc0
+
+  mov   $r1, %rax
+  mulq  $acc1
+  add   $acc0, $t2
+  adc   \$0, %rdx
+  add   %rax, $t2
+  adc   \$0, %rdx
+
+  mov   $r1, $t3
+  imul  $acc2, $t3
+  add   %rdx, $t3
+############################
+
+  mov   $t0, $acc0
+  mov   $t1, $acc1
+  mov   $t2, $acc2
+  and   \$3, $acc2
+
+  mov   $t2, $t0
+  mov   $t3, $t1
+
+  and   \$-4, $t0
+  shrd  \$2, $t3, $t2
+  shr   \$2, $t3
+
+  add   $t0, $acc0
+  adc   $t1, $acc1
+  adc   \$0, $acc2
+
+  add   $t2, $acc0
+  adc   $t3, $acc1
+  adc   \$0, $acc2
+
+  sub   \$16, $inl
+  cmp   \$16, $inl
+  jae   1b
+
+2:
+  test  $inl, $inl
+  jz    3f
+
+  mov   \$1, $t0
+  xor   $t1, $t1
+  xor   $t2, $t2
+  add   $inl, $inp
+
+4:
+  shld  \$8, $t0, $t1
+  shl   \$8, $t0
+  movzxb  -1($inp), $t2
+  xor   $t2, $t0
+  dec   $inp
+  dec   $inl
+  jnz   4b
+
+  add   $t0, $acc0
+  adc   $t1, $acc1
+  adc   \$0, $acc2
+
+  mov   \$16, $inl
+  jmp   5b
+
+3:
+
+  mov $acc0, 8*0($state)
+  mov $acc1, 8*1($state)
+  mov $acc2, 8*2($state)
+
+  pop %r15
+  pop %r14
+  pop %r13
+  pop %r12
+  pop %r11
+  ret
+.size poly1305_update_x64, .-poly1305_update_x64
+___
+}
+
+{
+
+my ($mac, $state)=("%rsi", "%rdi");
+
+my ($acc0, $acc1, $acc2, $t0, $t1, $t2)
+   =("%rcx", "%rax", "%rdx", "%r8", "%r9", "%r10");
+
+$code.=<<___;
+###############################################################################
+# void poly1305_finish_x64(void* state, uint64_t mac[2]);
+.type poly1305_finish_x64,\@function, 2
+.align 64
+.globl poly1305_finish_x64
+poly1305_finish_x64:
+
+  mov 8*0($state), $acc0
+  mov 8*1($state), $acc1
+  mov 8*2($state), $acc2
+
+  mov $acc0, $t0
+  mov $acc1, $t1
+  mov $acc2, $t2
+
+  sub \$-5, $acc0
+  sbb \$-1, $acc1
+  sbb \$3, $acc2
+
+  cmovc $t0, $acc0
+  cmovc $t1, $acc1
+  cmovc $t2, $acc2
+
+  add 8*5($state), $acc0
+  adc 8*6($state), $acc1
+  mov $acc0, ($mac)
+  mov $acc1, 8($mac)
+
+  ret
+.size poly1305_finish_x64, .-poly1305_finish_x64
+___
+}
+}
+$code =~ s/\`([^\`]*)\`/eval($1)/gem;
+print $code;
+close STDOUT;
diff --git a/crypto/chacha20poly1305/chacha20.c b/crypto/chacha20poly1305/chacha20.c
new file mode 100644
index 0000000..3044751
--- /dev/null
+++ b/crypto/chacha20poly1305/chacha20.c
@@ -0,0 +1,162 @@
+/* Copyright (c) 2014, Google Inc.
+ *
+ * Permission to use, copy, modify, and/or distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
+ * SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
+ * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
+ * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */
+
+/* Adapted from the public domain, estream code by D. Bernstein. */
+
+#include "chacha20poly1305.h"
+
+/* sigma contains the ChaCha constants, which happen to be an ASCII string. */
+static const char sigma[16] = "expand 32-byte k";
+
+#define ROTATE(v, n) (((v) << (n)) | ((v) >> (32 - (n))))
+#define XOR(v, w) ((v) ^ (w))
+#define PLUS(x, y) ((x) + (y))
+#define PLUSONE(v) (PLUS((v), 1))
+
+#define U32TO8_LITTLE(p, v)    \
+  {                            \
+    (p)[0] = (v >> 0) & 0xff;  \
+    (p)[1] = (v >> 8) & 0xff;  \
+    (p)[2] = (v >> 16) & 0xff; \
+    (p)[3] = (v >> 24) & 0xff; \
+  }
+
+#define U8TO32_LITTLE(p)                              \
+  (((uint32_t)((p)[0])) | ((uint32_t)((p)[1]) << 8) | \
+   ((uint32_t)((p)[2]) << 16) | ((uint32_t)((p)[3]) << 24))
+
+/* QUARTERROUND updates a, b, c, d with a ChaCha "quarter" round. */
+#define QUARTERROUND(a,b,c,d) \
+  x[a] = PLUS(x[a],x[b]); x[d] = ROTATE(XOR(x[d],x[a]),16); \
+  x[c] = PLUS(x[c],x[d]); x[b] = ROTATE(XOR(x[b],x[c]),12); \
+  x[a] = PLUS(x[a],x[b]); x[d] = ROTATE(XOR(x[d],x[a]), 8); \
+  x[c] = PLUS(x[c],x[d]); x[b] = ROTATE(XOR(x[b],x[c]), 7);
+
+/* chacha_core performs |num_rounds| rounds of ChaCha20 on the input words in
+ * |input| and writes the 64 output bytes to |output|. */
+static void chacha_core(uint8_t output[64], const uint32_t input[16]) {
+  uint32_t x[16];
+  int i;
+
+  memcpy(x, input, sizeof(uint32_t) * 16);
+  for (i = 20; i > 0; i -= 2) {
+    QUARTERROUND(0, 4, 8, 12)
+    QUARTERROUND(1, 5, 9, 13)
+    QUARTERROUND(2, 6, 10, 14)
+    QUARTERROUND(3, 7, 11, 15)
+    QUARTERROUND(0, 5, 10, 15)
+    QUARTERROUND(1, 6, 11, 12)
+    QUARTERROUND(2, 7, 8, 13)
+    QUARTERROUND(3, 4, 9, 14)
+  }
+
+  for (i = 0; i < 16; ++i) {
+    x[i] = PLUS(x[i], input[i]);
+  }
+  for (i = 0; i < 16; ++i) {
+    U32TO8_LITTLE(output + 4 * i, x[i]);
+  }
+}
+
+void CRYPTO_chacha_20(uint8_t *out, const uint8_t *in, size_t in_len,
+                      uint8_t nonce[48]) {
+
+#ifdef CHAPOLY_x86_64_ASM
+  const int AVX2_threshold = 256;
+  const int AVX2_min_buf = 128;
+  const int AVX_min_buf = 64;
+
+  uint8_t buf[128];
+  size_t buf_size;
+
+  void (*core_func)(uint8_t *out, const uint8_t *in, size_t in_len,
+                    uint8_t nonce[48]) = NULL;
+#else /* !CHAPOLY_x86_64_ASM */
+
+  uint8_t buf[64];
+
+#endif
+
+  uint32_t input[16];
+  size_t todo, i;
+
+#ifdef CHAPOLY_x86_64_ASM
+  if (in_len >= AVX2_threshold && ((OPENSSL_ia32cap_loc()[1] >> 5) & 1)) {
+    buf_size = AVX2_min_buf;
+    core_func = chacha_20_core_avx2;
+  } else if ((OPENSSL_ia32cap_loc()[0] >> 60) & 1) {
+    buf_size = AVX_min_buf;
+    core_func = chacha_20_core_avx;
+  } else goto do_legacy;
+
+  core_func(out, in, in_len, nonce);
+  todo = in_len & (buf_size - 1);
+
+  if(todo) {
+    out += in_len - todo;
+    in += in_len - todo;
+    memcpy(buf, in, todo);
+
+    core_func(buf, buf, buf_size, nonce);
+
+    memcpy(out, buf, todo);
+    memset(buf, 0, buf_size);
+  }
+  return;
+
+do_legacy:
+#endif /* CHAPOLY_x86_64_ASM */
+
+  input[0] = U8TO32_LITTLE(sigma + 0);
+  input[1] = U8TO32_LITTLE(sigma + 4);
+  input[2] = U8TO32_LITTLE(sigma + 8);
+  input[3] = U8TO32_LITTLE(sigma + 12);
+
+  input[4] = U8TO32_LITTLE(nonce + 0);
+  input[5] = U8TO32_LITTLE(nonce + 4);
+  input[6] = U8TO32_LITTLE(nonce + 8);
+  input[7] = U8TO32_LITTLE(nonce + 12);
+
+  input[8] = U8TO32_LITTLE(nonce + 16);
+  input[9] = U8TO32_LITTLE(nonce + 20);
+  input[10] = U8TO32_LITTLE(nonce + 24);
+  input[11] = U8TO32_LITTLE(nonce + 28);
+
+  input[12] = U8TO32_LITTLE(nonce + 32);
+  input[13] = U8TO32_LITTLE(nonce + 36);
+  input[14] = U8TO32_LITTLE(nonce + 40);
+  input[15] = U8TO32_LITTLE(nonce + 44);
+
+  while (in_len > 0) {
+    todo = 64;
+    if (in_len < todo) {
+      todo = in_len;
+    }
+
+    chacha_core(buf, input);
+    for (i = 0; i < todo; i++) {
+      out[i] = in[i] ^ buf[i];
+    }
+
+    out += todo;
+    in += todo;
+    in_len -= todo;
+
+    ((uint64_t*)input)[6]++;
+  }
+
+  U32TO8_LITTLE(nonce + 32, input[12]);
+  U32TO8_LITTLE(nonce + 36, input[13]);
+}
+
diff --git a/crypto/chacha20poly1305/chacha20poly1305.h b/crypto/chacha20poly1305/chacha20poly1305.h
new file mode 100644
index 0000000..09b0450
--- /dev/null
+++ b/crypto/chacha20poly1305/chacha20poly1305.h
@@ -0,0 +1,79 @@
+/* Copyright (c) 2014, Google Inc.
+ *
+ * Permission to use, copy, modify, and/or distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
+ * SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
+ * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
+ * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */
+
+#ifndef OPENSSL_HEADER_POLY1305_H
+#define OPENSSL_HEADER_POLY1305_H
+
+#include <stdint.h>
+#include <stddef.h>
+#include <string.h>
+#include "crypto.h"
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+#define POLY1305_MAC_LEN  (16)
+#define POLY1305_PAD_LEN  (16)
+
+typedef unsigned char poly1305_state[372];
+
+
+/* CRYPTO_poly1305_init sets up |state| so that it can be used to calculate an
+ * authentication tag with the one-time key |key|. Note that |key| is a
+ * one-time key and therefore there is no `reset' method because that would
+ * enable several messages to be authenticated with the same key. */
+void CRYPTO_poly1305_init(poly1305_state* state, const uint8_t key[32]);
+
+/* CRYPTO_poly1305_update processes |in_len| bytes from |in|. It can be called
+ * zero or more times after poly1305_init. */
+void CRYPTO_poly1305_update(poly1305_state* state, const uint8_t* in,
+                            size_t in_len);
+
+/* CRYPTO_poly1305_finish completes the poly1305 calculation and writes a 16
+ * byte authentication tag to |mac|. */
+void CRYPTO_poly1305_finish(poly1305_state* state,
+                            uint8_t mac[POLY1305_MAC_LEN]);
+
+/* CRYPTO_chacha_20 encrypts |in_len| bytes from |in| with the given key and
+ * nonce and writes the result to |out|, which may be equal to |in|. The
+ * initial block counter is specified by |counter|. */
+void CRYPTO_chacha_20(uint8_t *out, const uint8_t *in, size_t in_len,
+                      uint8_t nonce[48]);
+
+#ifdef CHAPOLY_x86_64_ASM
+void poly1305_init_x64(poly1305_state* state, const uint8_t key[32]);
+void poly1305_update_x64(poly1305_state* state, const uint8_t *in, size_t in_len);
+void poly1305_finish_x64(poly1305_state* state, uint8_t mac[16]);
+
+void poly1305_init_avx(poly1305_state* state, const uint8_t key[32]);
+void poly1305_update_avx(poly1305_state* state, const uint8_t *in, size_t in_len);
+void poly1305_finish_avx(poly1305_state* state, uint8_t mac[16]);
+
+void poly1305_update_avx2(poly1305_state* state, const uint8_t *in, size_t in_len);
+void poly1305_finish_avx2(poly1305_state* state, uint8_t mac[16]);
+
+void chacha_20_core_avx(uint8_t *out, const uint8_t *in, size_t in_len,
+                        uint8_t nonce[48]);
+
+void chacha_20_core_avx2(uint8_t *out, const uint8_t *in, size_t in_len,
+                         uint8_t nonce[48]);
+#endif
+
+
+#if defined(__cplusplus)
+}  /* extern C */
+#endif
+
+#endif  /* OPENSSL_HEADER_POLY1305_H */
diff --git a/crypto/chacha20poly1305/chapolytest.c b/crypto/chacha20poly1305/chapolytest.c
new file mode 100644
index 0000000..7e2933f
--- /dev/null
+++ b/crypto/chacha20poly1305/chapolytest.c
@@ -0,0 +1,470 @@
+/* ====================================================================
+ * Copyright (c) 2011-2013 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.OpenSSL.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    licensing@OpenSSL.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.OpenSSL.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ */
+
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+
+#include <openssl/chacha20poly1305.h>
+
+struct chacha_test {
+    const char *noncehex;
+    const char *outhex;
+};
+
+struct poly1305_test {
+    const char *inputhex;
+    const char *keyhex;
+    const char *outhex;
+};
+
+static const struct chacha_test chacha_tests[] = {
+    {
+        "00000000000000000000000000000000""00000000000000000000000000000000"
+        "00000000000000000000000000000000",
+        "76b8e0ada0f13d90405d6ae55386bd28""bdd219b8a08ded1aa836efcc8b770dc7"
+        "da41597c5157488d7724e03fb8d84a37""6a43b8f41518a11cc387b669b2ee6586",
+    },
+    {
+        "00000000000000000000000000000000""00000000000000000000000000000001"
+        "00000000000000000000000000000000",
+        "4540f05a9f1fb296d7736e7b208e3c96""eb4fe1834688d2604f450952ed432d41"
+        "bbe2a0b6ea7566d2a5d1e7e20d42af2c""53d792b1c43fea817e9ad275ae546963",
+    },
+    {
+        "00000000000000000000000000000000""00000000000000000000000000000000"
+        "00000000000000000000000000000001",
+        "de9cba7bf3d69ef5e786dc63973f653a""0b49e015adbff7134fcb7df137821031"
+        "e85a050278a7084527214f73efc7fa5b""5277062eb7a0433e445f41e31afab757",
+    },
+    {
+        "00000000000000000000000000000000""00000000000000000000000000000000"
+        "00000000000000000100000000000000",
+        "ef3fdfd6c61578fbf5cf35bd3dd33b80""09631634d21e42ac33960bd138e50d32"
+        "111e4caf237ee53ca8ad6426194a8854""5ddc497a0b466e7d6bbdb0041b2f586b",
+    },
+    {
+        "000102030405060708090a0b0c0d0e0f""101112131415161718191a1b1c1d1e1f"
+        "00000000000000000001020304050607",
+        "f798a189f195e66982105ffb640bb775""7f579da31602fc93ec01ac56f85ac3c1"
+        "34a4547b733b46413042c94400491769""05d3be59ea1c53f15916155c2be8241a"
+        "38008b9a26bc35941e2444177c8ade66""89de95264986d95889fb60e84629c9bd"
+        "9a5acb1cc118be563eb9b3a4a472f82e""09a7e778492b562ef7130e88dfe031c7"
+        "9db9d4f7c7a899151b9a475032b63fc3""85245fe054e3dd5a97a5f576fe064025"
+        "d3ce042c566ab2c507b138db853e3d69""59660996546cc9c4a6eafdc777c040d7"
+        "0eaf46f76dad3979e5c5360c3317166a""1c894c94a371876a94df7628fe4eaaf2"
+        "ccb27d5aaae0ad7ad0f9d4b6ad3b5409""8746d4524d38407a6deb",
+    },
+};
+
+static const struct poly1305_test poly1305_tests[] = {
+    {
+        "",
+        "c8afaac331ee372cd6082de134943b17""4710130e9f6fea8d72293850a667d86c",
+        "4710130e9f6fea8d72293850a667d86c",
+    },
+    {
+        "48656c6c6f20776f726c6421",
+        "746869732069732033322d6279746520""6b657920666f7220506f6c7931333035",
+        "a6f745008f81c916a20dcc74eef2b2f0",
+    },
+    {
+        "00000000000000000000000000000000""00000000000000000000000000000000",
+        "746869732069732033322d6279746520""6b657920666f7220506f6c7931333035",
+        "49ec78090e481ec6c26b33b91ccc0307",
+    },
+    {
+        "43727970746f6772617068696320466f""72756d2052657365617263682047726f"
+        "7570",
+        "85d6be7857556d337f4452fe42d506a8""0103808afb0db2fd4abff6af4149f51b",
+        "a8061dc1305136c6c22b8baf0c0127a9"
+    },
+    {
+        "f3f6",
+        "851fc40c3467ac0be05cc20404f3f700""580b3b0f9447bb1e69d095b5928b6dbc",
+        "f4c633c3044fc145f84f335cb81953de"
+    },
+    {
+        "",
+        "a0f3080000f46400d0c7e9076c834403""dd3fab2251f11ac759f0887129cc2ee7",
+        "dd3fab2251f11ac759f0887129cc2ee7"
+    },
+    {
+        "663cea190ffb83d89593f3f476b6bc24""d7e679107ea26adb8caf6652d0656136",
+        "48443d0bb0d21109c89a100b5ce2c208""83149c69b561dd88298a1798b10716ef",
+        "0ee1c16bb73f0f4fd19881753c01cdbe"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d1""36c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67""fa83e158c994d961c4cb21095c1bf9",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "5154ad0d2cb26e01274fc51148491f1b"
+    },
+    /*
+     * self-generated
+     */
+    {
+        "ab0812724a7f1e342742cbed374d94d1""36c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67""fa83e158c994d961c4cb21095c1bf9af",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "812059a5da198637cac7c4a631bee466"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d1""36c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "5b88d7f6228b11e2e28579a5c0c1f761"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "bbb613b2b6d753ba07395b916aaece15"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "c794d7057d1778c4bbee0a39b3d97342"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "ffbcb9b371423152d7fca5ad042fbaa9"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136"
+        "812059a5da198637cac7c4a631bee466",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "069ed6b8ef0f207b3e243bb1019fe632"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136"
+        "812059a5da198637cac7c4a631bee4665b88d7f6228b11e2e28579a5c0c1f761",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "cca339d9a45fa2368c2c68b3a4179133"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136"
+        "812059a5da198637cac7c4a631bee4665b88d7f6228b11e2e28579a5c0c1f761"
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "53f6e828a2f0fe0ee815bf0bd5841a34"
+    },
+    {
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136"
+        "812059a5da198637cac7c4a631bee4665b88d7f6228b11e2e28579a5c0c1f761"
+        "ab0812724a7f1e342742cbed374d94d136c6b8795d45b3819830f2c04491faf0"
+        "990c62e48b8018b2c3e4a0fa3134cb67fa83e158c994d961c4cb21095c1bf9af"
+        "48443d0bb0d21109c89a100b5ce2c20883149c69b561dd88298a1798b10716ef"
+        "663cea190ffb83d89593f3f476b6bc24d7e679107ea26adb8caf6652d0656136"
+        "812059a5da198637cac7c4a631bee4665b88d7f6228b11e2e28579a5c0c1f761",
+        "12976a08c4426d0ce8a82407c4f48207""80f8c20aa71202d1e29179cbcb555a57",
+        "b846d44e9bbd53cedffbfbb6b7fa4933"
+    },
+    {
+    /*
+        * poly1305_ieee754.c failed this in final stage
+        */
+        "842364e156336c0998b933a6237726180d9e3fdcbde4cd5d17080fc3beb49614"
+        "d7122c037463ff104d73f19c12704628d417c4c54a3fe30d3c3d7714382d43b0"
+        "382a50a5dee54be844b076e8df88201a1cd43b90eb21643fa96f39b518aa8340"
+        "c942ff3c31baf7c9bdbf0f31ae3fa096bf8c63030609829fe72e179824890bc8"
+        "e08c315c1cce2a83144dbbff09f74e3efc770b54d0984a8f19b14719e6363564"
+        "1d6b1eedf63efbf080e1783d32445412114c20de0b837a0dfa33d6b82825fff4"
+        "4c9a70ea54ce47f07df698e6b03323b53079364a5fc3e9dd034392bdde86dccd"
+        "da94321c5e44060489336cb65bf3989c36f7282c2f5d2b882c171e74",
+        "95d5c005503e510d8cd0aa072c4a4d06""6eabc52d11653df47fbf63ab198bcc26",
+        "f248312e578d9d58f8b7bb4d19105431"
+    },
+    /*
+        * test vectors from Google
+        */
+    {
+        "",
+        "c8afaac331ee372cd6082de134943b17""4710130e9f6fea8d72293850a667d86c",
+        "4710130e9f6fea8d72293850a667d86c",
+    },
+    {
+        "48656c6c6f20776f726c6421",
+        "746869732069732033322d6279746520""6b657920666f7220506f6c7931333035",
+        "a6f745008f81c916a20dcc74eef2b2f0"
+    },
+    {
+        "0000000000000000000000000000000000000000000000000000000000000000",
+        "746869732069732033322d6279746520""6b657920666f7220506f6c7931333035",
+        "49ec78090e481ec6c26b33b91ccc0307"
+    },
+    /*
+        * test vectors from Andrew Moon
+        */
+    {   /* nacl */
+        "8e993b9f48681273c29650ba32fc76ce48332ea7164d96a4476fb8c531a1186a"
+        "c0dfc17c98dce87b4da7f011ec48c97271d2c20f9b928fe2270d6fb863d51738"
+        "b48eeee314a7cc8ab932164548e526ae90224368517acfeabd6bb3732bc0e9da"
+        "99832b61ca01b6de56244a9e88d5f9b37973f622a43d14a6599b1f654cb45a74"
+        "e355a5",
+        "eea6a7251c1e72916d11c2cb214d3c25""2539121d8e234e652d651fa4c8cff880",
+        "f3ffc7703f9400e52a7dfb4b3d3305d9"
+    },
+    {   /* wrap 2^130-5 */
+        "ffffffffffffffffffffffffffffffff",
+        "02000000000000000000000000000000""00000000000000000000000000000000",
+        "03000000000000000000000000000000"
+    },
+    {   /* wrap 2^128 */
+        "02000000000000000000000000000000",
+        "02000000000000000000000000000000""ffffffffffffffffffffffffffffffff",
+        "03000000000000000000000000000000"
+    },
+    {   /* limb carry */
+        "fffffffffffffffffffffffffffffffff0ffffffffffffffffffffffffffffff"
+        "11000000000000000000000000000000",
+        "01000000000000000000000000000000""00000000000000000000000000000000",
+        "05000000000000000000000000000000"
+    },
+    {   /* 2^130-5 */
+        "fffffffffffffffffffffffffffffffffbfefefefefefefefefefefefefefefe"
+        "01010101010101010101010101010101",
+        "01000000000000000000000000000000""00000000000000000000000000000000",
+        "00000000000000000000000000000000"
+    },
+    {   /* 2^130-6 */
+        "fdffffffffffffffffffffffffffffff",
+        "02000000000000000000000000000000""00000000000000000000000000000000",
+        "faffffffffffffffffffffffffffffff"
+    },
+    {   /* 5*H+L reduction intermediate */
+        "e33594d7505e43b900000000000000003394d7505e4379cd0100000000000000"
+        "0000000000000000000000000000000001000000000000000000000000000000",
+        "01000000000000000400000000000000""00000000000000000000000000000000",
+        "14000000000000005500000000000000"
+    },
+    {   /* 5*H+L reduction final */
+        "e33594d7505e43b900000000000000003394d7505e4379cd0100000000000000"
+        "00000000000000000000000000000000",
+        "01000000000000000400000000000000""00000000000000000000000000000000",
+        "13000000000000000000000000000000"
+    }
+};
+
+static unsigned char hex_digit(char h)
+{
+    if (h >= '0' && h <= '9')
+        return h - '0';
+    else if (h >= 'a' && h <= 'f')
+        return h - 'a' + 10;
+    else if (h >= 'A' && h <= 'F')
+        return h - 'A' + 10;
+    else
+        abort();
+}
+
+static void hex_decode(unsigned char *out, const char* hex)
+{
+    size_t j = 0;
+
+    while (*hex != 0) {
+        unsigned char v = hex_digit(*hex++);
+        v <<= 4;
+        v |= hex_digit(*hex++);
+        out[j++] = v;
+    }
+}
+
+static void hexdump(unsigned char *a, size_t len)
+{
+    size_t i;
+
+    for (i = 0; i < len; i++) {
+        printf("%02x", a[i]);
+    }
+}
+
+/* misalign returns a pointer that points 0 to 15 bytes into |in| such that the
+ * returned pointer has alignment 1 mod 16. */
+static void* misalign(void* in)
+{
+    intptr_t x = (intptr_t) in;
+    x += (17 - (x % 16)) % 16;
+    return (void*) x;
+}
+
+int main()
+{
+
+    unsigned num_tests =
+        sizeof(chacha_tests) / sizeof(struct chacha_test);
+    unsigned i;
+    unsigned char nonce_bytes[48 + 16] = {0};
+
+    for (i = 0; i < num_tests; i++) {
+        unsigned char *nonce = misalign(nonce_bytes);
+
+        printf("ChaCha20 test #%d\n", i);
+        const struct chacha_test *test = &chacha_tests[i];
+        unsigned char *expected, *out_bytes, *zero_bytes, *out, *zeros;
+        size_t len = strlen(test->outhex);
+
+        if (strlen(test->noncehex) != 48*2 || (len & 1) == 1)
+            return 1;
+
+        len /= 2;
+
+        hex_decode(nonce, test->noncehex);
+
+        expected = malloc(len);
+        out_bytes = malloc(len+16);
+        zero_bytes = malloc(len+16);
+        /* Attempt to test unaligned inputs. */
+        out = misalign(out_bytes);
+        zeros = misalign(zero_bytes);
+        memset(zeros, 0, len);
+
+        hex_decode(expected, test->outhex);
+        CRYPTO_chacha_20(out, zeros, len, nonce);
+
+        if (memcmp(out, expected, len) != 0) {
+            printf("ChaCha20 test #%d failed.\n", i);
+            printf("got:      ");
+            hexdump(out, len);
+            printf("\nexpected: ");
+            hexdump(expected, len);
+            printf("\n");
+            return 1;
+        }
+
+
+        free(expected);
+        free(zero_bytes);
+        free(out_bytes);
+    }
+
+    num_tests =
+        sizeof(poly1305_tests) / sizeof(struct poly1305_test);
+    unsigned char key[32], out[16], expected[16];
+    poly1305_state poly1305;
+
+    for (i = 0; i < num_tests; i++) {
+        printf("Poly1305 test #%d\n", i);
+        const struct poly1305_test *test = &poly1305_tests[i];
+        unsigned char *in;
+        size_t inlen = strlen(test->inputhex);
+
+        if (strlen(test->keyhex) != sizeof(key)*2 ||
+            strlen(test->outhex) != sizeof(out)*2 ||
+            (inlen & 1) == 1)
+            return 1;
+
+        inlen /= 2;
+
+        hex_decode(key, test->keyhex);
+        hex_decode(expected, test->outhex);
+
+        in = malloc(inlen);
+
+        hex_decode(in, test->inputhex);
+
+#ifdef CHAPOLY_x86_64_ASM
+        if((OPENSSL_ia32cap_loc()[1] >> 5) & 1) {
+            poly1305_init_x64(&poly1305, key);
+            poly1305_update_avx2(&poly1305, in, inlen);
+            poly1305_finish_avx2(&poly1305, out);
+        } else {
+            poly1305_init_x64(&poly1305, key);
+            poly1305_update_x64(&poly1305, in, inlen);
+            poly1305_finish_x64(&poly1305, out);
+        } 
+#else
+        {
+            CRYPTO_poly1305_init(&poly1305, key);
+            CRYPTO_poly1305_update(&poly1305, in, inlen);
+            CRYPTO_poly1305_finish(&poly1305, out);
+        }
+#endif
+        if (memcmp(out, expected, sizeof(expected)) != 0) {
+            printf("Poly1305 test #%d failed.\n", i);
+            printf("got:      ");
+            hexdump(out, sizeof(out));
+            printf("\nexpected: ");
+            hexdump(expected, sizeof(expected));
+            printf("\n");
+            return 1;
+        }
+
+        free(in);
+    }
+
+    printf("PASS\n");
+    return 0;
+}
+
diff --git a/crypto/chacha20poly1305/poly1305.c b/crypto/chacha20poly1305/poly1305.c
new file mode 100644
index 0000000..50bc4a0
--- /dev/null
+++ b/crypto/chacha20poly1305/poly1305.c
@@ -0,0 +1,287 @@
+/* Copyright (c) 2014, Google Inc.
+ *
+ * Permission to use, copy, modify, and/or distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
+ * SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
+ * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
+ * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */
+
+/* This implementation of poly1305 is by Andrew Moon
+ * (https://github.com/floodyberry/poly1305-donna) and released as public
+ * domain. */
+
+#include "chacha20poly1305.h"
+
+#include <string.h>
+
+#if !defined(B_ENDIAN) 
+/* We can assume little-endian. */
+static uint32_t U8TO32_LE(const uint8_t *m) {
+  uint32_t r;
+  memcpy(&r, m, sizeof(r));
+  return r;
+}
+
+static void U32TO8_LE(uint8_t *m, uint32_t v) { memcpy(m, &v, sizeof(v)); }
+#else
+static uint32_t U8TO32_LE(const uint8_t *m) {
+  return (uint32_t)m[0] | (uint32_t)m[1] << 8 | (uint32_t)m[2] << 16 |
+         (uint32_t)m[3] << 24;
+}
+
+static void U32TO8_LE(uint8_t *m, uint32_t v) {
+  m[0] = v;
+  m[1] = v >> 8;
+  m[2] = v >> 16;
+  m[3] = v >> 24;
+}
+#endif
+
+static uint64_t mul32x32_64(uint32_t a, uint32_t b) { return (uint64_t)a * b; }
+
+struct poly1305_state_st {
+  uint32_t r0, r1, r2, r3, r4;
+  uint32_t s1, s2, s3, s4;
+  uint32_t h0, h1, h2, h3, h4;
+  uint8_t buf[16];
+  unsigned int buf_used;
+  uint8_t key[16];
+};
+
+/* poly1305_blocks updates |state| given some amount of input data. This
+ * function may only be called with a |len| that is not a multiple of 16 at the
+ * end of the data. Otherwise the input must be buffered into 16 byte blocks. */
+static void poly1305_update(struct poly1305_state_st *state, const uint8_t *in,
+                            size_t len) {
+  uint32_t t0, t1, t2, t3;
+  uint64_t t[5];
+  uint32_t b;
+  uint64_t c;
+  size_t j;
+  uint8_t mp[16];
+
+  if (len < 16) {
+    goto poly1305_donna_atmost15bytes;
+  }
+
+poly1305_donna_16bytes:
+  t0 = U8TO32_LE(in);
+  t1 = U8TO32_LE(in + 4);
+  t2 = U8TO32_LE(in + 8);
+  t3 = U8TO32_LE(in + 12);
+
+  in += 16;
+  len -= 16;
+
+  state->h0 += t0 & 0x3ffffff;
+  state->h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff;
+  state->h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff;
+  state->h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff;
+  state->h4 += (t3 >> 8) | (1 << 24);
+
+poly1305_donna_mul:
+  t[0] = mul32x32_64(state->h0, state->r0) + mul32x32_64(state->h1, state->s4) +
+         mul32x32_64(state->h2, state->s3) + mul32x32_64(state->h3, state->s2) +
+         mul32x32_64(state->h4, state->s1);
+  t[1] = mul32x32_64(state->h0, state->r1) + mul32x32_64(state->h1, state->r0) +
+         mul32x32_64(state->h2, state->s4) + mul32x32_64(state->h3, state->s3) +
+         mul32x32_64(state->h4, state->s2);
+  t[2] = mul32x32_64(state->h0, state->r2) + mul32x32_64(state->h1, state->r1) +
+         mul32x32_64(state->h2, state->r0) + mul32x32_64(state->h3, state->s4) +
+         mul32x32_64(state->h4, state->s3);
+  t[3] = mul32x32_64(state->h0, state->r3) + mul32x32_64(state->h1, state->r2) +
+         mul32x32_64(state->h2, state->r1) + mul32x32_64(state->h3, state->r0) +
+         mul32x32_64(state->h4, state->s4);
+  t[4] = mul32x32_64(state->h0, state->r4) + mul32x32_64(state->h1, state->r3) +
+         mul32x32_64(state->h2, state->r2) + mul32x32_64(state->h3, state->r1) +
+         mul32x32_64(state->h4, state->r0);
+
+  state->h0 = (uint32_t)t[0] & 0x3ffffff;
+  c = (t[0] >> 26);
+  t[1] += c;
+  state->h1 = (uint32_t)t[1] & 0x3ffffff;
+  b = (uint32_t)(t[1] >> 26);
+  t[2] += b;
+  state->h2 = (uint32_t)t[2] & 0x3ffffff;
+  b = (uint32_t)(t[2] >> 26);
+  t[3] += b;
+  state->h3 = (uint32_t)t[3] & 0x3ffffff;
+  b = (uint32_t)(t[3] >> 26);
+  t[4] += b;
+  state->h4 = (uint32_t)t[4] & 0x3ffffff;
+  b = (uint32_t)(t[4] >> 26);
+  state->h0 += b * 5;
+
+  if (len >= 16)
+    goto poly1305_donna_16bytes;
+
+/* final bytes */
+poly1305_donna_atmost15bytes:
+  if (!len)
+    return;
+
+  for (j = 0; j < len; j++)
+    mp[j] = in[j];
+  mp[j++] = 1;
+  for (; j < 16; j++)
+    mp[j] = 0;
+  len = 0;
+
+  t0 = U8TO32_LE(mp + 0);
+  t1 = U8TO32_LE(mp + 4);
+  t2 = U8TO32_LE(mp + 8);
+  t3 = U8TO32_LE(mp + 12);
+
+  state->h0 += t0 & 0x3ffffff;
+  state->h1 += ((((uint64_t)t1 << 32) | t0) >> 26) & 0x3ffffff;
+  state->h2 += ((((uint64_t)t2 << 32) | t1) >> 20) & 0x3ffffff;
+  state->h3 += ((((uint64_t)t3 << 32) | t2) >> 14) & 0x3ffffff;
+  state->h4 += (t3 >> 8);
+
+  goto poly1305_donna_mul;
+}
+
+void CRYPTO_poly1305_init(poly1305_state *statep, const uint8_t key[32]) {
+  struct poly1305_state_st *state = (struct poly1305_state_st *)statep;
+  uint32_t t0, t1, t2, t3;
+
+  t0 = U8TO32_LE(key + 0);
+  t1 = U8TO32_LE(key + 4);
+  t2 = U8TO32_LE(key + 8);
+  t3 = U8TO32_LE(key + 12);
+
+  /* precompute multipliers */
+  state->r0 = t0 & 0x3ffffff;
+  t0 >>= 26;
+  t0 |= t1 << 6;
+  state->r1 = t0 & 0x3ffff03;
+  t1 >>= 20;
+  t1 |= t2 << 12;
+  state->r2 = t1 & 0x3ffc0ff;
+  t2 >>= 14;
+  t2 |= t3 << 18;
+  state->r3 = t2 & 0x3f03fff;
+  t3 >>= 8;
+  state->r4 = t3 & 0x00fffff;
+
+  state->s1 = state->r1 * 5;
+  state->s2 = state->r2 * 5;
+  state->s3 = state->r3 * 5;
+  state->s4 = state->r4 * 5;
+
+  /* init state */
+  state->h0 = 0;
+  state->h1 = 0;
+  state->h2 = 0;
+  state->h3 = 0;
+  state->h4 = 0;
+
+  state->buf_used = 0;
+  memcpy(state->key, key + 16, sizeof(state->key));
+}
+
+void CRYPTO_poly1305_update(poly1305_state *statep, const uint8_t *in,
+                            size_t in_len) {
+  unsigned int i;
+  struct poly1305_state_st *state = (struct poly1305_state_st *)statep;
+
+  if (state->buf_used) {
+    unsigned int todo = 16 - state->buf_used;
+    if (todo > in_len)
+      todo = in_len;
+    for (i = 0; i < todo; i++)
+      state->buf[state->buf_used + i] = in[i];
+    state->buf_used += todo;
+    in_len -= todo;
+    in += todo;
+
+    if (state->buf_used == 16) {
+      poly1305_update(state, state->buf, 16);
+      state->buf_used = 0;
+    }
+  }
+
+  if (in_len >= 16) {
+    size_t todo = in_len & ~0xf;
+    poly1305_update(state, in, todo);
+    in += todo;
+    in_len &= 0xf;
+  }
+
+  if (in_len) {
+    for (i = 0; i < in_len; i++)
+      state->buf[i] = in[i];
+    state->buf_used = in_len;
+  }
+}
+
+void CRYPTO_poly1305_finish(poly1305_state *statep, uint8_t mac[16]) {
+  struct poly1305_state_st *state = (struct poly1305_state_st *)statep;
+  uint64_t f0, f1, f2, f3;
+  uint32_t g0, g1, g2, g3, g4;
+  uint32_t b, nb;
+
+  if (state->buf_used)
+    poly1305_update(state, state->buf, state->buf_used);
+
+  b = state->h0 >> 26;
+  state->h0 = state->h0 & 0x3ffffff;
+  state->h1 += b;
+  b = state->h1 >> 26;
+  state->h1 = state->h1 & 0x3ffffff;
+  state->h2 += b;
+  b = state->h2 >> 26;
+  state->h2 = state->h2 & 0x3ffffff;
+  state->h3 += b;
+  b = state->h3 >> 26;
+  state->h3 = state->h3 & 0x3ffffff;
+  state->h4 += b;
+  b = state->h4 >> 26;
+  state->h4 = state->h4 & 0x3ffffff;
+  state->h0 += b * 5;
+
+  g0 = state->h0 + 5;
+  b = g0 >> 26;
+  g0 &= 0x3ffffff;
+  g1 = state->h1 + b;
+  b = g1 >> 26;
+  g1 &= 0x3ffffff;
+  g2 = state->h2 + b;
+  b = g2 >> 26;
+  g2 &= 0x3ffffff;
+  g3 = state->h3 + b;
+  b = g3 >> 26;
+  g3 &= 0x3ffffff;
+  g4 = state->h4 + b - (1 << 26);
+
+  b = (g4 >> 31) - 1;
+  nb = ~b;
+  state->h0 = (state->h0 & nb) | (g0 & b);
+  state->h1 = (state->h1 & nb) | (g1 & b);
+  state->h2 = (state->h2 & nb) | (g2 & b);
+  state->h3 = (state->h3 & nb) | (g3 & b);
+  state->h4 = (state->h4 & nb) | (g4 & b);
+
+  f0 = ((state->h0) | (state->h1 << 26)) + (uint64_t)U8TO32_LE(&state->key[0]);
+  f1 = ((state->h1 >> 6) | (state->h2 << 20)) +
+       (uint64_t)U8TO32_LE(&state->key[4]);
+  f2 = ((state->h2 >> 12) | (state->h3 << 14)) +
+       (uint64_t)U8TO32_LE(&state->key[8]);
+  f3 = ((state->h3 >> 18) | (state->h4 << 8)) +
+       (uint64_t)U8TO32_LE(&state->key[12]);
+
+  U32TO8_LE(&mac[0], f0);
+  f1 += (f0 >> 32);
+  U32TO8_LE(&mac[4], f1);
+  f2 += (f1 >> 32);
+  U32TO8_LE(&mac[8], f2);
+  f3 += (f2 >> 32);
+  U32TO8_LE(&mac[12], f3);
+}
+
diff --git a/crypto/cryptlib.c b/crypto/cryptlib.c
index c9f674b..c6ff6ff 100644
--- a/crypto/cryptlib.c
+++ b/crypto/cryptlib.c
@@ -656,16 +656,6 @@ const char *CRYPTO_get_lock_name(int type)
 extern unsigned int OPENSSL_ia32cap_P[4];
 unsigned long *OPENSSL_ia32cap_loc(void)
 {
-    if (sizeof(long) == 4)
-        /*
-         * If 32-bit application pulls address of OPENSSL_ia32cap_P[0]
-         * clear second element to maintain the illusion that vector
-         * is 32-bit.
-         */
-        OPENSSL_ia32cap_P[1] = 0;
-
-    OPENSSL_ia32cap_P[2] = 0;
-
     return (unsigned long *)OPENSSL_ia32cap_P;
 }
 
diff --git a/crypto/evp/Makefile b/crypto/evp/Makefile
index aaaad98..e30b588 100644
--- a/crypto/evp/Makefile
+++ b/crypto/evp/Makefile
@@ -29,7 +29,8 @@ LIBSRC= encode.c digest.c evp_enc.c evp_key.c evp_acnf.c evp_cnf.c \
 	c_all.c c_allc.c c_alld.c evp_lib.c bio_ok.c \
 	evp_pkey.c evp_pbe.c p5_crpt.c p5_crpt2.c \
 	e_old.c pmeth_lib.c pmeth_fn.c pmeth_gn.c m_sigver.c \
-	e_aes_cbc_hmac_sha1.c e_aes_cbc_hmac_sha256.c e_rc4_hmac_md5.c
+	e_aes_cbc_hmac_sha1.c e_aes_cbc_hmac_sha256.c e_rc4_hmac_md5.c \
+	e_chacha20poly1305.c
 
 LIBOBJ=	encode.o digest.o evp_enc.o evp_key.o evp_acnf.o evp_cnf.o \
 	e_des.o e_bf.o e_idea.o e_des3.o e_camellia.o\
@@ -42,7 +43,8 @@ LIBOBJ=	encode.o digest.o evp_enc.o evp_key.o evp_acnf.o evp_cnf.o \
 	c_all.o c_allc.o c_alld.o evp_lib.o bio_ok.o \
 	evp_pkey.o evp_pbe.o p5_crpt.o p5_crpt2.o \
 	e_old.o pmeth_lib.o pmeth_fn.o pmeth_gn.o m_sigver.o \
-	e_aes_cbc_hmac_sha1.o e_aes_cbc_hmac_sha256.o e_rc4_hmac_md5.o
+	e_aes_cbc_hmac_sha1.o e_aes_cbc_hmac_sha256.o e_rc4_hmac_md5.o \
+	e_chacha20poly1305.o
 
 SRC= $(LIBSRC)
 
@@ -263,6 +265,7 @@ e_cast.o: ../../include/openssl/objects.h ../../include/openssl/opensslconf.h
 e_cast.o: ../../include/openssl/opensslv.h ../../include/openssl/ossl_typ.h
 e_cast.o: ../../include/openssl/safestack.h ../../include/openssl/stack.h
 e_cast.o: ../../include/openssl/symhacks.h ../cryptlib.h e_cast.c evp_locl.h
+e_chacha20poly1305.o: ../../include/openssl/chacha20poly1305.h e_chacha20poly1305.c
 e_des.o: ../../e_os.h ../../include/openssl/asn1.h ../../include/openssl/bio.h
 e_des.o: ../../include/openssl/buffer.h ../../include/openssl/crypto.h
 e_des.o: ../../include/openssl/des.h ../../include/openssl/des_old.h
diff --git a/crypto/evp/e_chacha20poly1305.c b/crypto/evp/e_chacha20poly1305.c
new file mode 100644
index 0000000..17f8cb4
--- /dev/null
+++ b/crypto/evp/e_chacha20poly1305.c
@@ -0,0 +1,435 @@
+/* ====================================================================
+ * Copyright (c) 2001-2014 The OpenSSL Project.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in
+ *    the documentation and/or other materials provided with the
+ *    distribution.
+ *
+ * 3. All advertising materials mentioning features or use of this
+ *    software must display the following acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
+ *
+ * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
+ *    endorse or promote products derived from this software without
+ *    prior written permission. For written permission, please contact
+ *    openssl-core@openssl.org.
+ *
+ * 5. Products derived from this software may not be called "OpenSSL"
+ *    nor may "OpenSSL" appear in their names without prior written
+ *    permission of the OpenSSL Project.
+ *
+ * 6. Redistributions of any form whatsoever must retain the following
+ *    acknowledgment:
+ *    "This product includes software developed by the OpenSSL Project
+ *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
+ * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
+ * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ * OF THE POSSIBILITY OF SUCH DAMAGE.
+ * ====================================================================
+ *
+ */
+
+#include <openssl/opensslconf.h>
+#ifndef OPENSSL_NO_CHACHA_POLY
+#include <openssl/evp.h>
+#include <openssl/err.h>
+#include <openssl/chacha20poly1305.h>
+#include "evp_locl.h"
+#include <openssl/rand.h>
+
+#define FILL_BUFFER ((size_t)128)
+
+typedef struct {
+    uint8_t        iv[12];
+    uint8_t        nonce[48];
+    size_t         aad_l;
+    size_t         ct_l;
+    unsigned       valid:1;
+    unsigned       draft:1;
+    uint8_t        poly_buffer[FILL_BUFFER];
+    uint8_t        chacha_buffer[FILL_BUFFER];
+    uint16_t       poly_buffer_used;
+    uint16_t       chacha_used;
+#ifdef CHAPOLY_x86_64_ASM
+    void (*poly1305_init_ptr)(poly1305_state *, const uint8_t *);
+    void (*poly1305_update_ptr)(poly1305_state *, const uint8_t *, size_t);
+    void (*poly1305_finish_ptr)(poly1305_state *, uint8_t *);
+    poly1305_state poly_state;
+    #define poly_init aead_ctx->poly1305_init_ptr
+    #define poly_update poly1305_update_wrapper
+    #define poly_finish poly1305_finish_wrapper
+#else
+    #define poly_init CRYPTO_poly1305_init
+    #define poly_update(c,i,l) CRYPTO_poly1305_update(&c->poly_state,i,l)
+    #define poly_finish(c,m) CRYPTO_poly1305_finish(&c->poly_state,m)
+#endif
+} EVP_CHACHA20_POLY1305_CTX;
+
+
+#ifdef CHAPOLY_x86_64_ASM
+#include <immintrin.h>
+
+static void poly1305_update_wrapper(EVP_CHACHA20_POLY1305_CTX *ctx,
+                                    const uint8_t *in,
+                                    size_t in_len)
+{
+    int todo;
+    /* Attempt to fill as many bytes as possible before calling the update
+       function */
+    if (in_len < FILL_BUFFER || ctx->poly_buffer_used) {
+        todo = FILL_BUFFER - ctx->poly_buffer_used;
+        todo = in_len < todo? in_len : todo;
+        memcpy(ctx->poly_buffer + ctx->poly_buffer_used, in, todo);
+        ctx->poly_buffer_used += todo;
+        in += todo;
+        in_len -= todo;
+
+        if (ctx->poly_buffer_used == FILL_BUFFER) {
+            ctx->poly1305_update_ptr(&ctx->poly_state,
+                                     ctx->poly_buffer,
+                                     FILL_BUFFER);
+            ctx->poly_buffer_used = 0;
+        }
+    }
+
+    if (in_len >= FILL_BUFFER) {
+        ctx->poly1305_update_ptr(&ctx->poly_state, in, in_len & (-FILL_BUFFER));
+        in += in_len & (-FILL_BUFFER);
+        in_len &= (FILL_BUFFER - 1);
+    }
+
+    if (in_len) {
+        memcpy(ctx->poly_buffer, in, in_len);
+        ctx->poly_buffer_used = in_len;
+    }
+}
+
+
+static void poly1305_finish_wrapper(EVP_CHACHA20_POLY1305_CTX *ctx,
+                                    uint8_t mac[POLY1305_MAC_LEN])
+{
+    if (ctx->poly_buffer_used) {
+
+        if (ctx->poly_buffer_used % POLY1305_PAD_LEN) {
+            memset(ctx->poly_buffer + ctx->poly_buffer_used, 0,
+            POLY1305_PAD_LEN - (ctx->poly_buffer_used % POLY1305_PAD_LEN));
+        }
+
+        ctx->poly1305_update_ptr(&ctx->poly_state,
+                                 ctx->poly_buffer,
+                                 ctx->poly_buffer_used);
+    }
+
+    ctx->poly1305_finish_ptr(&ctx->poly_state, mac);
+    memset(ctx->poly_buffer, 0, FILL_BUFFER);
+}
+#endif
+
+
+#ifdef CHAPOLY_x86_64_ASM
+static void EVP_chacha20_poly1305_cpuid(EVP_CHACHA20_POLY1305_CTX *ctx)
+{
+    if ((OPENSSL_ia32cap_loc()[1] >> 5) & 1) {          /* AVX2 */
+        ctx->poly1305_init_ptr = poly1305_init_x64;     /* Lazy init */
+        ctx->poly1305_update_ptr = poly1305_update_avx2;
+        ctx->poly1305_finish_ptr = poly1305_finish_avx2;
+/*
+    } else if (0 && (OPENSSL_ia32cap_loc()[0] >> 60) & 1) {  // AVX -disabled
+        ctx->poly1305_init_ptr = poly1305_init_avx;
+        ctx->poly1305_update_ptr = poly1305_update_avx;
+        ctx->poly1305_finish_ptr = poly1305_finish_avx;
+*/
+    } else {                                            /* x64 code */
+        ctx->poly1305_init_ptr = poly1305_init_x64;
+        ctx->poly1305_update_ptr = poly1305_update_x64;
+        ctx->poly1305_finish_ptr = poly1305_finish_x64;
+    }
+}
+#endif
+
+
+static int EVP_chacha20_poly1305_init_draft(EVP_CIPHER_CTX *ctx,
+                                            const unsigned char *key,
+                                            const unsigned char *iv,
+                                            int enc)
+{
+    EVP_CHACHA20_POLY1305_CTX *aead_ctx = ctx->cipher_data;
+    memcpy(aead_ctx->nonce, key, 32);
+    aead_ctx->valid = 0;
+    aead_ctx->draft = 1;
+
+#ifdef CHAPOLY_x86_64_ASM
+    EVP_chacha20_poly1305_cpuid(aead_ctx);
+#endif
+
+    return 1;
+}
+
+
+static int EVP_chacha20_poly1305_init(EVP_CIPHER_CTX *ctx,
+                                      const unsigned char *key,
+                                      const unsigned char *iv,
+                                      int enc)
+{
+    EVP_CHACHA20_POLY1305_CTX *aead_ctx = ctx->cipher_data;
+    memcpy(aead_ctx->nonce, key, 32);
+    memcpy(aead_ctx->iv, iv, 12);
+    aead_ctx->valid = 0;
+    aead_ctx->draft = 0;
+
+#ifdef CHAPOLY_x86_64_ASM
+    EVP_chacha20_poly1305_cpuid(aead_ctx);
+#endif
+
+    return 1;
+}
+
+
+static int EVP_chacha20_poly1305_cipher(EVP_CIPHER_CTX *ctx,
+                                        unsigned char *out,
+                                        const unsigned char *in,
+                                        size_t inl)
+{
+    EVP_CHACHA20_POLY1305_CTX *aead_ctx = ctx->cipher_data;
+    uint8_t  poly_mac[POLY1305_MAC_LEN];
+    uint8_t  zero[POLY1305_PAD_LEN] = {0};
+    uint64_t cmp;
+    int      i, todo;
+
+    if (!aead_ctx->valid)
+        return 0;
+
+    if (inl < POLY1305_MAC_LEN)
+        return -1;
+
+    /* Fix for MAC */
+    inl -= POLY1305_MAC_LEN;
+
+    if (!ctx->encrypt) {
+        poly_update(aead_ctx, in, inl);
+    }
+
+    i = 0;
+    if (inl < 256) {
+        /* Consume the buffer we computed during poly initialization */
+        todo = inl > (FILL_BUFFER - aead_ctx->chacha_used) ?
+               FILL_BUFFER - aead_ctx->chacha_used :
+               inl;
+
+#ifdef CHAPOLY_x86_64_ASM
+        for (; i <= todo - 16; i+=16) {
+            _mm_storeu_si128((__m128i*)&out[i],
+                  _mm_xor_si128(_mm_loadu_si128((__m128i *)&in[i]),
+                     _mm_loadu_si128((__m128i *)&aead_ctx->chacha_buffer[i + 64])));
+        }
+#endif
+        for (; i < todo; i++) {
+            out[i] = in[i] ^ aead_ctx->chacha_buffer[i + 64 /*aead_ctx->chacha_used*/];
+        }
+
+    } else {
+        /* For long messages don't use precomputed buffer */
+        ((uint64_t *)(aead_ctx->nonce))[4]--;
+    }
+
+    todo = inl - i;
+
+    if (todo) {
+        CRYPTO_chacha_20(&out[i], &in[i], todo, aead_ctx->nonce);
+    }
+
+    if (ctx->encrypt) {
+        poly_update(aead_ctx, out, inl);
+    }
+
+    aead_ctx->ct_l += inl;
+
+    if (!aead_ctx->draft) {
+        /* For RFC padd ciphertext with zeroes, then mac len(aad)||len(ct) */
+        todo = aead_ctx->ct_l % POLY1305_PAD_LEN ?
+               POLY1305_PAD_LEN - (aead_ctx->ct_l % POLY1305_PAD_LEN) :
+               0;
+
+        if (todo) {
+            poly_update(aead_ctx, zero, todo);
+        }
+
+        poly_update(aead_ctx, (uint8_t*)&aead_ctx->aad_l, sizeof(uint64_t));
+        poly_update(aead_ctx, (uint8_t*)&aead_ctx->ct_l, sizeof(uint64_t));
+
+    } else {
+        /* For the draft don't pad, mac len(ct) */
+        poly_update(aead_ctx, (uint8_t*)&aead_ctx->ct_l, sizeof(uint64_t));
+    }
+    aead_ctx->valid = 0;
+
+    if (ctx->encrypt) {
+        poly_finish(aead_ctx, &out[inl]);
+        return inl + POLY1305_MAC_LEN;
+
+    } else { /* Decryption */
+        poly_finish(aead_ctx, poly_mac);
+        /* Constant time comparison */
+        cmp = (*(uint64_t *)(poly_mac)) ^ (*(uint64_t *)(in + inl));
+        cmp |= (*(uint64_t *)(poly_mac + 8)) ^ (*(uint64_t *)(in + inl + 8));
+
+        if (cmp) {
+            OPENSSL_cleanse(out, inl);
+            return -1;
+        }
+
+        return inl;
+    }
+}
+
+
+static int EVP_chacha20_poly1305_cleanup(EVP_CIPHER_CTX *ctx)
+{
+    return 1;
+}
+
+
+static int EVP_chacha20_poly1305_ctrl(EVP_CIPHER_CTX *ctx,
+                                      int type,
+                                      int arg,
+                                      void *ptr)
+{
+    EVP_CHACHA20_POLY1305_CTX *aead_ctx = ctx->cipher_data;
+    uint8_t        aad[EVP_AEAD_TLS1_AAD_LEN + 8];
+    uint64_t       thirteen = EVP_AEAD_TLS1_AAD_LEN;
+
+    switch (type) {
+        case EVP_CTRL_AEAD_TLS1_AAD:
+
+            if (arg != EVP_AEAD_TLS1_AAD_LEN)
+                return 0;
+
+            /* Initialize poly keys */
+            memset(aead_ctx->chacha_buffer, 0, FILL_BUFFER);
+
+            if (!aead_ctx->draft) {
+                /* RFC IV = (0 || iv) ^ seq_num */
+                memset(aead_ctx->nonce + 32, 0, 4);
+                memcpy(aead_ctx->nonce + 36, aead_ctx->iv, 12);
+                *(uint64_t *)(aead_ctx->nonce + 40) ^= *(uint64_t *)(ptr);
+
+            } else {
+                /* draft IV = 0 || seq_num */
+                memset(aead_ctx->nonce + 32, 0, 8);
+                memcpy(aead_ctx->nonce + 40, ptr, 8);
+            }
+            /* Poly keys = ENC(0) */
+            CRYPTO_chacha_20(aead_ctx->chacha_buffer,
+                             aead_ctx->chacha_buffer,
+                             FILL_BUFFER,
+                             aead_ctx->nonce);
+
+            poly_init(&aead_ctx->poly_state, aead_ctx->chacha_buffer);
+
+            aead_ctx->chacha_used = 64;
+            aead_ctx->poly_buffer_used = 0;
+            aead_ctx->aad_l = arg;
+            aead_ctx->ct_l = 0;
+
+            /* Absorb AAD */
+            memcpy(aad, ptr, arg);
+            memset(aad + arg, 0, sizeof(aad) - arg);
+
+            /* If decrypting fix length for tag */
+            if (!ctx->encrypt) {
+                unsigned int len = (aad[arg-2] << 8) | aad[arg-1];
+                len -= POLY1305_MAC_LEN;
+                aad[arg-2] = len>>8;
+                aad[arg-1] = len & 0xff;
+            }
+
+            if (!aead_ctx->draft) {
+                /* In the RFC, AAD is padded with zeroes */
+                poly_update(aead_ctx, aad, POLY1305_PAD_LEN);
+
+            } else {
+                /* In the draft AAD is followed by len(AAD) */
+                memcpy(&aad[arg], &thirteen, sizeof(thirteen));
+                poly_update(aead_ctx, aad, arg + sizeof(thirteen));
+            }
+
+            aead_ctx->valid = 1;
+            return POLY1305_MAC_LEN;
+
+            break;
+
+        default:
+            return 0;
+            break;
+    }
+
+    return 0;
+}
+
+
+#define CUSTOM_FLAGS    (\
+          EVP_CIPH_CUSTOM_IV | EVP_CIPH_FLAG_CUSTOM_CIPHER \
+        | EVP_CIPH_ALWAYS_CALL_INIT  \
+        | EVP_CIPH_CUSTOM_COPY)
+
+
+static const EVP_CIPHER chacha20_poly1305_d = {
+    0,    /* nid ??? */
+    1,    /* block size, sorta */
+    32,   /* key len */
+    0,    /* iv len */
+    CUSTOM_FLAGS|EVP_CIPH_FLAG_AEAD_CIPHER,    /* flags */
+    EVP_chacha20_poly1305_init_draft,
+    EVP_chacha20_poly1305_cipher,
+    EVP_chacha20_poly1305_cleanup,
+    sizeof(EVP_CHACHA20_POLY1305_CTX),         /* ctx size */
+    NULL,
+    NULL,
+    EVP_chacha20_poly1305_ctrl,
+    NULL
+    };
+
+
+static const EVP_CIPHER chacha20_poly1305 = {
+    0,    /* nid ??? */
+    1,    /* block size, sorta */
+    32,   /* key len */
+    12,   /* iv len */
+    CUSTOM_FLAGS|EVP_CIPH_FLAG_AEAD_CIPHER,    /* flags */
+    EVP_chacha20_poly1305_init,
+    EVP_chacha20_poly1305_cipher,
+    EVP_chacha20_poly1305_cleanup,
+    sizeof(EVP_CHACHA20_POLY1305_CTX),         /* ctx size */
+    NULL,
+    NULL,
+    EVP_chacha20_poly1305_ctrl,
+    NULL
+    };
+
+
+const EVP_CIPHER *EVP_chacha20_poly1305_draft(void)
+{ return &chacha20_poly1305_d; }
+
+
+const EVP_CIPHER *EVP_chacha20_poly1305(void)
+{ return &chacha20_poly1305; }
+#endif
diff --git a/crypto/evp/evp.h b/crypto/evp/evp.h
index 39ab793..53ed671 100644
--- a/crypto/evp/evp.h
+++ b/crypto/evp/evp.h
@@ -893,6 +893,10 @@ const EVP_CIPHER *EVP_camellia_256_cfb128(void);
 #  define EVP_camellia_256_cfb EVP_camellia_256_cfb128
 const EVP_CIPHER *EVP_camellia_256_ofb(void);
 # endif
+# ifndef OPENSSL_NO_CHACHA_POLY
+const EVP_CIPHER *EVP_chacha20_poly1305(void);
+const EVP_CIPHER *EVP_chacha20_poly1305_draft(void);
+# endif
 
 # ifndef OPENSSL_NO_SEED
 const EVP_CIPHER *EVP_seed_ecb(void);
diff --git a/ssl/s3_lib.c b/ssl/s3_lib.c
index f846cb5..b0364c1 100644
--- a/ssl/s3_lib.c
+++ b/ssl/s3_lib.c
@@ -2891,6 +2891,111 @@ OPENSSL_GLOBAL SSL_CIPHER ssl3_ciphers[] = {
      256},
 #endif
 
+#if !defined(OPENSSL_NO_CHACHA_POLY)
+     /* Draft ciphers */
+    {
+     1,
+     TLS1_TXT_ECDHE_RSA_WITH_CHACHA20_POLY1305_D,
+     TLS1_CK_ECDHE_RSA_WITH_CHACHA20_POLY1305_D,
+     SSL_kEECDH,
+     SSL_aRSA,
+     SSL_CHACHA20POLY1305_D,
+     SSL_AEAD,
+     SSL_TLSV1_2,
+     SSL_HIGH,
+     SSL_HANDSHAKE_MAC_SHA256 | TLS1_PRF_SHA256,
+     256,
+     256},
+    {
+     1,
+     TLS1_TXT_ECDHE_ECDSA_WITH_CHACHA20_POLY1305_D,
+     TLS1_CK_ECDHE_ECDSA_WITH_CHACHA20_POLY1305_D,
+     SSL_kEECDH,
+     SSL_aECDSA,
+     SSL_CHACHA20POLY1305_D,
+     SSL_AEAD,
+     SSL_TLSV1_2,
+     SSL_HIGH,
+     SSL_HANDSHAKE_MAC_SHA256 | TLS1_PRF_SHA256,
+     256,
+     256},
+    {
+     1,
+     TLS1_TXT_DHE_RSA_WITH_CHACHA20_POLY1305_D,
+     TLS1_CK_DHE_RSA_WITH_CHACHA20_POLY1305_D,
+     SSL_kEDH,
+     SSL_aRSA,
+     SSL_CHACHA20POLY1305_D,
+     SSL_AEAD,
+     SSL_TLSV1_2,
+     SSL_HIGH,
+     SSL_HANDSHAKE_MAC_SHA256 | TLS1_PRF_SHA256,
+     256,
+     256},
+    /* RFC ciphers */
+    /* Cipher CCA8 as per draft-ietf-tls-chacha20-poly1305-03 */
+    {
+     1,
+     TLS1_TXT_ECDHE_RSA_WITH_CHACHA20_POLY1305,
+     TLS1_CK_ECDHE_RSA_WITH_CHACHA20_POLY1305,
+     SSL_kECDHE,
+     SSL_aRSA,
+     SSL_CHACHA20POLY1305,
+     SSL_AEAD,
+     SSL_TLSV1_2,
+     SSL_HIGH,
+     SSL_HANDSHAKE_MAC_SHA256 | TLS1_PRF_SHA256,
+     256,
+     256,
+     },
+    /* Cipher CCA9 */
+    {
+     1,
+     TLS1_TXT_ECDHE_ECDSA_WITH_CHACHA20_POLY1305,
+     TLS1_CK_ECDHE_ECDSA_WITH_CHACHA20_POLY1305,
+     SSL_kECDHE,
+     SSL_aECDSA,
+     SSL_CHACHA20POLY1305,
+     SSL_AEAD,
+     SSL_TLSV1_2,
+     SSL_HIGH,
+     SSL_HANDSHAKE_MAC_SHA256 | TLS1_PRF_SHA256,
+     256,
+     256,
+     },
+    /* Cipher CCAA */
+    {
+     1,
+     TLS1_TXT_DHE_RSA_WITH_CHACHA20_POLY1305,
+     TLS1_CK_DHE_RSA_WITH_CHACHA20_POLY1305,
+     SSL_kDHE,
+     SSL_aRSA,
+     SSL_CHACHA20POLY1305,
+     SSL_AEAD,
+     SSL_TLSV1_2,
+     SSL_HIGH,
+     SSL_HANDSHAKE_MAC_SHA256 | TLS1_PRF_SHA256,
+     256,
+     256,
+     },
+    /* Cipher CCAB */
+    {
+     1,
+     TLS1_TXT_PSK_WITH_CHACHA20_POLY1305,
+     TLS1_CK_PSK_WITH_CHACHA20_POLY1305,
+     SSL_kPSK,
+     SSL_aPSK,
+     SSL_CHACHA20POLY1305,
+     SSL_AEAD,
+     SSL_TLSV1_2,
+     SSL_HIGH,
+     SSL_HANDSHAKE_MAC_SHA256 | TLS1_PRF_SHA256,
+     256,
+     256,
+     },
+#endif
+
+
 /* end of list */
 };
 
@@ -4036,6 +4141,7 @@ SSL_CIPHER *ssl3_choose_cipher(SSL *s, STACK_OF(SSL_CIPHER) *clnt,
     int i, ii, ok;
     CERT *cert;
     unsigned long alg_k, alg_a, mask_k, mask_a, emask_k, emask_a;
+    int use_chacha = 0;
 
     /* Let's see which ciphers we can support */
     cert = s->cert;
@@ -4069,9 +4175,17 @@ SSL_CIPHER *ssl3_choose_cipher(SSL *s, STACK_OF(SSL_CIPHER) *clnt,
     if (s->options & SSL_OP_CIPHER_SERVER_PREFERENCE || tls1_suiteb(s)) {
         prio = srvr;
         allow = clnt;
+        /* Use ChaCha20+Poly1305 iff it's client's most preferred cipher suite */
+        if (sk_SSL_CIPHER_num(clnt) > 0) {
+            c = sk_SSL_CIPHER_value(clnt, 0);
+            if (c->algorithm_enc == SSL_CHACHA20POLY1305 ||
+                c->algorithm_enc == SSL_CHACHA20POLY1305_D)
+                use_chacha = 1;
+        }
     } else {
         prio = clnt;
         allow = srvr;
+        use_chacha = 1;
     }
 
     tls1_set_cert_validity(s);
@@ -4083,6 +4197,11 @@ SSL_CIPHER *ssl3_choose_cipher(SSL *s, STACK_OF(SSL_CIPHER) *clnt,
         if ((c->algorithm_ssl & SSL_TLSV1_2) && !SSL_USE_TLS1_2_CIPHERS(s))
             continue;
 
+        /* Skip ChaCha unless top client priority */
+        if ((c->algorithm_enc == SSL_CHACHA20POLY1305 ||
+             c->algorithm_enc == SSL_CHACHA20POLY1305_D) && !use_chacha)
+            continue;
+
         ssl_set_cert_masks(cert, c);
         mask_k = cert->mask_k;
         mask_a = cert->mask_a;
diff --git a/ssl/ssl.h b/ssl/ssl.h
index ae8c925..c7635b6 100644
--- a/ssl/ssl.h
+++ b/ssl/ssl.h
@@ -294,6 +294,8 @@ extern "C" {
 # define SSL_TXT_AES256          "AES256"
 # define SSL_TXT_AES             "AES"
 # define SSL_TXT_AES_GCM         "AESGCM"
+# define SSL_TXT_CHACHA20_D      "CHACHA20-draft"
+# define SSL_TXT_CHACHA20        "CHACHA20"
 # define SSL_TXT_CAMELLIA128     "CAMELLIA128"
 # define SSL_TXT_CAMELLIA256     "CAMELLIA256"
 # define SSL_TXT_CAMELLIA        "CAMELLIA"
diff --git a/ssl/ssl_ciph.c b/ssl/ssl_ciph.c
index 6957bda..21eae5c 100644
--- a/ssl/ssl_ciph.c
+++ b/ssl/ssl_ciph.c
@@ -150,25 +150,27 @@
 #endif
 #include "ssl_locl.h"
 
-#define SSL_ENC_DES_IDX         0
-#define SSL_ENC_3DES_IDX        1
-#define SSL_ENC_RC4_IDX         2
-#define SSL_ENC_RC2_IDX         3
-#define SSL_ENC_IDEA_IDX        4
-#define SSL_ENC_NULL_IDX        5
-#define SSL_ENC_AES128_IDX      6
-#define SSL_ENC_AES256_IDX      7
-#define SSL_ENC_CAMELLIA128_IDX 8
-#define SSL_ENC_CAMELLIA256_IDX 9
-#define SSL_ENC_GOST89_IDX      10
-#define SSL_ENC_SEED_IDX        11
-#define SSL_ENC_AES128GCM_IDX   12
-#define SSL_ENC_AES256GCM_IDX   13
-#define SSL_ENC_NUM_IDX         14
+#define SSL_ENC_DES_IDX                     0
+#define SSL_ENC_3DES_IDX                    1
+#define SSL_ENC_RC4_IDX                     2
+#define SSL_ENC_RC2_IDX                     3
+#define SSL_ENC_IDEA_IDX                    4
+#define SSL_ENC_NULL_IDX                    5
+#define SSL_ENC_AES128_IDX                  6
+#define SSL_ENC_AES256_IDX                  7
+#define SSL_ENC_CAMELLIA128_IDX             8
+#define SSL_ENC_CAMELLIA256_IDX             9
+#define SSL_ENC_GOST89_IDX                  10
+#define SSL_ENC_SEED_IDX                    11
+#define SSL_ENC_AES128GCM_IDX               12
+#define SSL_ENC_AES256GCM_IDX               13
+#define SSL_ENC_CHACHA20POLY1305_DRAFT_IDX  14
+#define SSL_ENC_CHACHA20POLY1305_IDX        15
+#define SSL_ENC_NUM_IDX                     16
 
 static const EVP_CIPHER *ssl_cipher_methods[SSL_ENC_NUM_IDX] = {
     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
-    NULL, NULL
+    NULL, NULL, NULL, NULL
 };
 
 #define SSL_COMP_NULL_IDX       0
@@ -363,6 +365,9 @@ static const SSL_CIPHER cipher_aliases[] = {
     {0, SSL3_TXT_DHE_RSA_DES_192_CBC3_SHA, 0,
      SSL_kDHE, SSL_aRSA, SSL_3DES, SSL_SHA1, SSL_SSLV3,
      SSL_NOT_EXP | SSL_HIGH | SSL_FIPS, 0, 0, 0,},
+
+    {0, SSL_TXT_CHACHA20_D, 0, 0, 0, SSL_CHACHA20POLY1305_D, 0, 0, 0, 0, 0, 0},
+    {0, SSL_TXT_CHACHA20, 0, 0, 0, SSL_CHACHA20POLY1305, 0, 0, 0, 0, 0, 0},
 };
 
 /*
@@ -432,6 +437,11 @@ void ssl_load_ciphers(void)
     ssl_cipher_methods[SSL_ENC_AES256GCM_IDX] =
         EVP_get_cipherbyname(SN_aes_256_gcm);
 
+    ssl_cipher_methods[SSL_ENC_CHACHA20POLY1305_DRAFT_IDX] =
+        EVP_chacha20_poly1305_draft();
+    ssl_cipher_methods[SSL_ENC_CHACHA20POLY1305_IDX] =
+        EVP_chacha20_poly1305();
+
     ssl_digest_methods[SSL_MD_MD5_IDX] = EVP_get_digestbyname(SN_md5);
     ssl_mac_secret_size[SSL_MD_MD5_IDX] =
         EVP_MD_size(ssl_digest_methods[SSL_MD_MD5_IDX]);
@@ -582,6 +592,12 @@ int ssl_cipher_get_evp(const SSL_SESSION *s, const EVP_CIPHER **enc,
     case SSL_AES256GCM:
         i = SSL_ENC_AES256GCM_IDX;
         break;
+    case SSL_CHACHA20POLY1305_D:
+        i = SSL_ENC_CHACHA20POLY1305_DRAFT_IDX;
+        break;
+    case SSL_CHACHA20POLY1305:
+        i = SSL_ENC_CHACHA20POLY1305_IDX;
+        break;
     default:
         i = -1;
         break;
@@ -797,6 +813,12 @@ static void ssl_cipher_get_disabled(unsigned long *mkey, unsigned long *auth,
         (ssl_cipher_methods[SSL_ENC_AES256GCM_IDX] ==
          NULL) ? SSL_AES256GCM : 0;
     *enc |=
+        (ssl_cipher_methods[SSL_ENC_CHACHA20POLY1305_DRAFT_IDX] ==
+         NULL) ? SSL_CHACHA20POLY1305_D : 0;
+    *enc |=
+        (ssl_cipher_methods[SSL_ENC_CHACHA20POLY1305_IDX] ==
+         NULL) ? SSL_CHACHA20POLY1305 : 0;
+    *enc |=
         (ssl_cipher_methods[SSL_ENC_CAMELLIA128_IDX] ==
          NULL) ? SSL_CAMELLIA128 : 0;
     *enc |=
@@ -1812,6 +1834,12 @@ char *SSL_CIPHER_description(const SSL_CIPHER *cipher, char *buf, int len)
     case SSL_AES256GCM:
         enc = "AESGCM(256)";
         break;
+    case SSL_CHACHA20POLY1305_D:
+        enc = "ChaCha20-Poly1305-draft";
+        break;
+    case SSL_CHACHA20POLY1305:
+        enc = "ChaCha20-Poly1305";
+        break;
     case SSL_CAMELLIA128:
         enc = "Camellia(128)";
         break;
diff --git a/ssl/ssl_locl.h b/ssl/ssl_locl.h
index a8e4efc..67d355c 100644
--- a/ssl/ssl_locl.h
+++ b/ssl/ssl_locl.h
@@ -354,6 +354,8 @@
 # define SSL_SEED                0x00000800L
 # define SSL_AES128GCM           0x00001000L
 # define SSL_AES256GCM           0x00002000L
+# define SSL_CHACHA20POLY1305_D  0x00004000L
+# define SSL_CHACHA20POLY1305    0x00080000U  /* Value from openssl */
 
 # define SSL_AES                 (SSL_AES128|SSL_AES256|SSL_AES128GCM|SSL_AES256GCM)
 # define SSL_CAMELLIA            (SSL_CAMELLIA128|SSL_CAMELLIA256)
diff --git a/ssl/tls1.h b/ssl/tls1.h
index 7e237d0..fb0c981 100644
--- a/ssl/tls1.h
+++ b/ssl/tls1.h
@@ -563,6 +563,20 @@ SSL_CTX_callback_ctrl(ssl,SSL_CTRL_SET_TLSEXT_TICKET_KEY_CB,(void (*)(void))cb)
 # define TLS1_CK_ECDH_RSA_WITH_AES_128_GCM_SHA256        0x0300C031
 # define TLS1_CK_ECDH_RSA_WITH_AES_256_GCM_SHA384        0x0300C032
 
+/* ChaCha20-Poly1305 ciphersuites draft-agl-tls-chacha20poly1305-01 */
+# define TLS1_CK_ECDHE_RSA_WITH_CHACHA20_POLY1305_D      0x0300CC13
+# define TLS1_CK_ECDHE_ECDSA_WITH_CHACHA20_POLY1305_D    0x0300CC14
+# define TLS1_CK_DHE_RSA_WITH_CHACHA20_POLY1305_D        0x0300CC15
+
+/* ChaCha20-Poly1305 ciphersuites from RFC */
+# define TLS1_CK_ECDHE_RSA_WITH_CHACHA20_POLY1305        0x0300CCA8
+# define TLS1_CK_ECDHE_ECDSA_WITH_CHACHA20_POLY1305      0x0300CCA9
+# define TLS1_CK_DHE_RSA_WITH_CHACHA20_POLY1305          0x0300CCAA
+# define TLS1_CK_PSK_WITH_CHACHA20_POLY1305              0x0300CCAB
+# define TLS1_CK_ECDHE_PSK_WITH_CHACHA20_POLY1305        0x0300CCAC
+# define TLS1_CK_DHE_PSK_WITH_CHACHA20_POLY1305          0x0300CCAD
+# define TLS1_CK_RSA_PSK_WITH_CHACHA20_POLY1305          0x0300CCAE
+
 /*
  * XXX * Backward compatibility alert: + * Older versions of OpenSSL gave
  * some DHE ciphers names with "EDH" + * instead of "DHE".  Going forward, we
@@ -713,6 +727,20 @@ SSL_CTX_callback_ctrl(ssl,SSL_CTRL_SET_TLSEXT_TICKET_KEY_CB,(void (*)(void))cb)
 # define TLS1_TXT_ECDH_RSA_WITH_AES_128_GCM_SHA256       "ECDH-RSA-AES128-GCM-SHA256"
 # define TLS1_TXT_ECDH_RSA_WITH_AES_256_GCM_SHA384       "ECDH-RSA-AES256-GCM-SHA384"
 
+/* ChaCha20-Poly1305 ciphersuites draft-agl-tls-chacha20poly1305-01 */
+# define TLS1_TXT_ECDHE_RSA_WITH_CHACHA20_POLY1305_D     "ECDHE-RSA-CHACHA20-POLY1305-D"
+# define TLS1_TXT_ECDHE_ECDSA_WITH_CHACHA20_POLY1305_D   "ECDHE-ECDSA-CHACHA20-POLY1305-D"
+# define TLS1_TXT_DHE_RSA_WITH_CHACHA20_POLY1305_D       "DHE-RSA-CHACHA20-POLY1305-D"
+
+/* Chacha20-Poly1305 ciphersuites from RFC */
+# define TLS1_TXT_ECDHE_RSA_WITH_CHACHA20_POLY1305       "ECDHE-RSA-CHACHA20-POLY1305"
+# define TLS1_TXT_ECDHE_ECDSA_WITH_CHACHA20_POLY1305     "ECDHE-ECDSA-CHACHA20-POLY1305"
+# define TLS1_TXT_DHE_RSA_WITH_CHACHA20_POLY1305         "DHE-RSA-CHACHA20-POLY1305"
+# define TLS1_TXT_PSK_WITH_CHACHA20_POLY1305             "PSK-CHACHA20-POLY1305"
+# define TLS1_TXT_ECDHE_PSK_WITH_CHACHA20_POLY1305       "ECDHE-PSK-CHACHA20-POLY1305"
+# define TLS1_TXT_DHE_PSK_WITH_CHACHA20_POLY1305         "DHE-PSK-CHACHA20-POLY1305"
+# define TLS1_TXT_RSA_PSK_WITH_CHACHA20_POLY1305         "RSA-PSK-CHACHA20-POLY1305"
+
 # define TLS_CT_RSA_SIGN                 1
 # define TLS_CT_DSS_SIGN                 2
 # define TLS_CT_RSA_FIXED_DH             3
diff --git a/test/Makefile b/test/Makefile
index b180971..554d536 100644
--- a/test/Makefile
+++ b/test/Makefile
@@ -71,6 +71,7 @@
 VERIFYEXTRATEST=	verify_extra_test
 CLIENTHELLOTEST=	clienthellotest
 SSLV2CONFTEST = 	sslv2conftest
+CHAPOLYTEST=	chapolytest
 
 TESTS=		alltests
 
@@ -84,7 +85,7 @@
 	$(EVPTEST)$(EXE_EXT) $(EVPEXTRATEST)$(EXE_EXT) $(IGETEST)$(EXE_EXT) $(JPAKETEST)$(EXE_EXT) $(SRPTEST)$(EXE_EXT) \
 	$(ASN1TEST)$(EXE_EXT) $(V3NAMETEST)$(EXE_EXT) $(HEARTBEATTEST)$(EXE_EXT) \
 	$(CONSTTIMETEST)$(EXE_EXT) $(VERIFYEXTRATEST)$(EXE_EXT) \
-	$(CLIENTHELLOTEST)$(EXE_EXT) $(SSLV2CONFTEST)$(EXE_EXT)
+	$(CLIENTHELLOTEST)$(EXE_EXT) $(SSLV2CONFTEST)$(EXE_EXT) $(CHAPOLYTEST)$(EXE_EXT)
 
 # $(METHTEST)$(EXE_EXT)
 
@@ -98,7 +99,7 @@
 	$(BFTEST).o  $(SSLTEST).o  $(DSATEST).o  $(EXPTEST).o $(RSATEST).o \
 	$(EVPTEST).o $(EVPEXTRATEST).o $(IGETEST).o $(JPAKETEST).o $(ASN1TEST).o $(V3NAMETEST).o \
 	$(HEARTBEATTEST).o $(CONSTTIMETEST).o $(VERIFYEXTRATEST).o \
-	$(CLIENTHELLOTEST).o  $(SSLV2CONFTEST).o
+	$(CLIENTHELLOTEST).o  $(SSLV2CONFTEST).o $(CHAPOLYTEST).o
 
 SRC=	$(BNTEST).c $(ECTEST).c  $(ECDSATEST).c $(ECDHTEST).c $(IDEATEST).c \
 	$(MD2TEST).c  $(MD4TEST).c $(MD5TEST).c \
@@ -109,7 +110,7 @@
 	$(BFTEST).c  $(SSLTEST).c $(DSATEST).c   $(EXPTEST).c $(RSATEST).c \
 	$(EVPTEST).c $(EVPEXTRATEST).c $(IGETEST).c $(JPAKETEST).c $(SRPTEST).c $(ASN1TEST).c \
 	$(V3NAMETEST).c $(HEARTBEATTEST).c $(CONSTTIMETEST).c $(VERIFYEXTRATEST).c \
-	$(CLIENTHELLOTEST).c  $(SSLV2CONFTEST).c
+	$(CLIENTHELLOTEST).c  $(SSLV2CONFTEST).c $(CHAPOLYTEST).c
 
 EXHEADER= 
 HEADER=	testutil.h $(EXHEADER)
@@ -145,7 +146,7 @@
 	@(cd ..; $(MAKE) DIRS=apps all)
 
 alltests: \
-	test_des test_idea test_sha test_md4 test_md5 test_hmac \
+	test_des test_idea test_sha test_md4 test_md5 test_hmac test_chapoly \
 	test_md2 test_mdc2 test_wp \
 	test_rmd test_rc2 test_rc4 test_rc5 test_bf test_cast test_aes \
 	test_rand test_bn test_ec test_ecdsa test_ecdh \
@@ -366,6 +367,10 @@
 	@echo $(START) $@
 	../util/shlib_wrap.sh ./$(SSLV2CONFTEST)
 
+test_chapoly: $(CHAPOLYTEST)$(EXE_EXT)
+	@echo "Test ChaCha20 and Poly1305"
+	../util/shlib_wrap.sh ./$(CHAPOLYTEST)
+
 lint:
 	lint -DLINT $(INCLUDES) $(SRC)>fluff
 
@@ -546,6 +551,9 @@
 $(SSLV2CONFTEST)$(EXE_EXT): $(SSLV2CONFTEST).o
 	@target=$(SSLV2CONFTEST) $(BUILD_CMD)
 
+$(CHAPOLYTEST)$(EXE_EXT): $(CHAPOLYTEST).o
+	@target=$(CHAPOLYTEST); $(BUILD_CMD)
+
 #$(AESTEST).o: $(AESTEST).c
 #	$(CC) -c $(CFLAGS) -DINTERMEDIATE_VALUE_KAT -DTRACE_KAT_MCT $(AESTEST).c
 
@@ -614,6 +622,7 @@
 constant_time_test.o: ../crypto/constant_time_locl.h ../e_os.h
 constant_time_test.o: ../include/openssl/e_os2.h
 constant_time_test.o: ../include/openssl/opensslconf.h constant_time_test.c
+chapolytest.o: ../include/openssl/chacha20poly1305.h chapolytest.c
 destest.o: ../include/openssl/des.h ../include/openssl/des_old.h
 destest.o: ../include/openssl/e_os2.h ../include/openssl/opensslconf.h
 destest.o: ../include/openssl/ossl_typ.h ../include/openssl/safestack.h
-- 
2.5.0